Improving verbose queries using subset distribution.

Modeling reformulation using passage analysis.

Building a semantic representation for personal information.

Structural annotation of search queries using pseudo%2Drelevance feedback.

Building a semantic representation for personal information.

Geometric representations for multiple documents.

Evaluating verbose query processing techniques.

Ranking using multiple document types in desktop search.

Unsupervised estimation of dirichlet smoothing parameters.

Learning to rank query reformulations.

Query term ranking based on dependency parsing of verbose queries.

Learning concept importance using a weighted dependence model.
Modeling query concepts through term dependencies has been shown to have a significant positive effect on retrieval performance, especially for tasks such as web search, where relevance at high ranks is particularly critical. Most previous work, however, treats all concepts as equally important, an assumption that often does not hold, especially for longer, more complex queries. In this paper, we show that one of the most effective existing term dependence models can be naturally extended by assigning weights to concepts. We demonstrate that the weighted dependence model can be trained using existing learning-to-rank techniques, even with a relatively small number of training queries. Our study compares the effectiveness of both endogenous (collectionbased) and exogenous (based on external sources) features for determining concept importance. To test the weighted dependence model, we perform experiments on both publicly available TREC corpora and a proprietary web corpus. Our experimental results indicate that our model consistently and significantly outperforms both the standard bag-of-words model and the unweighted term dependence model, and that combining endogenous and exogenous features generally results in the best retrieval effectiveness.
Query reformulation using anchor text.
Query reformulation techniques based on query logs have been studied as a method of capturing user intent and improving retrieval effectiveness. The evaluation of these techniques has primarily, however, focused on proprietary query logs and selected samples of queries. In this paper, we suggest that anchor text, which is readily available, can be an effective substitute for a query log and study the effectiveness of a range of query reformulation techniques (including log-based stemming, substitution, and expansion) using standard TREC collections. Our results show that logbased query reformulation techniques are indeed effective with standard collections, but expansion is a much safer form of query modification than word substitution. We also show that using anchor text as a simulated query log is as least as effective as a real log for these techniques.
Retrieval experiments using pseudo%2Ddesktop collections.

Online community search using thread structure.

Query Evolution.
We show that XML views combined with XML query languages can provide surprisingly effective solutions to the problem of representing and querying the evolution of databases---both the evolution of their contents and the evolution of their schemas. Indeed, using XML, the histories of database relations can be represented naturally by means of temporally grouped data models. We show that schema changes and table histories that would be very difficult to represent in relational databases can be easily represented using XML, and also queried using XQuery.
A Probabilistic Retrieval Model for Semistructured Data.

Refining Keyword Queries for XML Retrieval by Combining Content and Structure.

Regression Rank Learning to Meet the Opportunity of Descriptive Queries.
and practitioners from academia and industry to discuss the latest developments in various aspects of online advertising and its interactions with information retrieval technologies. Our hope is that the
Transforming patents into prior%2Dart queries.

Two%2Dstage query segmentation for information retrieval.

Finding text reuse on the web.

Query structuring and expansion with two%2Dstage term dependence for Japanese web retrieval.

Unsolved problems in search %28and how we approach them%29.

Blog site search using resource selection.

A Statistical View of Binned Retrieval Models.

Evaluating Text Representations for Retrieval of the Best Group of Documents.

A cluster%2Dbased resampling method for pseudo%2Drelevance feedback.

Retrieval models for question and answer archives.

Discovering key concepts in verbose queries.
and practitioners from academia and industry to discuss the latest developments in various aspects of online advertising and its interactions with information retrieval technologies. Our hope is that the
Local text reuse detection.

UMass at TREC 2008 Blog Distillation Task.

An information%2Dpattern%2Dbased approach to novelty detection.

Hierarchical Language Models for Expert Finding in Enterprise Corpora.
Enterprise corpora contain evidence of what employees work on and therefore can be used to automatically find experts on a given topic. We present a general approach for representing the knowledge of a potential expert as a mixture of language models from associated documents. First we retrieve documents given the expert’s name using a generative probabilistic technique and weight the retrieved documents according to expert-specific posterior distribution. Then we model the expert indirectly through the set of associated documents, which allows us to exploit their underlying structure and complex language features. Experiments show that our method has excellent performance on TREC 2005 expert search task and that it effectively collects and combines evidence for expertise in a heterogeneous collection. 1
Measuring ranked list robustness for query performance prediction.

Introduction to the special issue on database and information retrieval integration.

Proximity%2Dbased document representation for named entity retrieval.

Web search from a bus.
Opportunistic connections to the Internet from open wireless access points is now commonly possible in municipal areas. Vehicular networks can opportunistically connect to the internet for several seconds via open access points. In this paper, we adapt the interactive process of web search and retrieval to vehicular networks with intermittent Internet access. Our system, called Thedu has mobile nodes use an Internet proxy to collect search engine results and prefetch result pages. The mobiles nodes download the pre-fetched web pages from the proxy. Our contribution is a novel set of techniques to make aggressive but selective prefetching practical, resulting in a significantly greater number of relevant web results returned to mobile users. To evaluate our scheme, we deployed Thedu on DieselNet, our vehicular testbed of buses operating in a micro-urban area around Amherst, MA. Using a simulated workload, we find that users can expect four times as many useful responses to web search queries compared to not using Thedu. Moreover, the mean latency in receiving the first relevant response for a query is 2.7 minutes when deployed in a semi-urban region with a sparser distribution of APs compared to big cities.
Modeling Term Associations for Ad%2DHoc Retrieval Performance Within Language Modeling Framework.


Investigating Retrieval Performance with Manually%2DBuilt Topic Models.
Modeling text with topics is currently a popular research area in both Machine Learning and Information Retrieval (IR). Most of this research has focused on automatic methods though there are many hand-crafted topic resources available online. In this paper we investigate retrieval performance with topic models constructed manually based on a hand-crafted directory resource. The original query is smoothed on the manually selected topic model, which can also be viewed as an “ideal ” user context model. Experiments with these topic models on the TREC retrieval tasks show that this type of topic model alone provides little benefit, and the overall performance is not as good as relevance modeling (which is an automatic query modification model). However, smoothing the query with topic models outperforms relevance models for a subset of the queries and automatic selection from these two models for particular queries gives better results overall than relevance models. We further demonstrate some improvements over relevance models with automatically built topic models based on the directory resource. 1.

Latent concept expansion using markov random fields.
Query expansion, in the form of pseudo-relevance feedback or relevance feedback, is a common technique used to improve retrieval effectiveness. Most previous approaches have ignored important issues, such as the role of features and the importance of modeling term dependencies. In this paper, we propose a robust query expansion technique based on the Markov random field model for information retrieval. The technique, called latent concept expansion, provides a mechanism for modeling term dependencies during expansion. Furthermore, the use of arbitrary features within the model provides a powerful framework for going beyond simple term occurrence features that are implicitly used by most other expansion techniques. We evaluate our technique against relevance models, a state-of-the-art language modeling query expansion technique. Our model demonstrates consistent and significant improvements in retrieval effectiveness across several TREC data sets. We also describe how our technique can be used to generate meaningful multi-term concepts for tasks such as query suggestion/reformulation.
Recommending citations for academic papers.
Abstract. Substantial effort is wasted in scientific circles by researchers who rediscover ideas that have already been published in the literature. This problem has been alleviated somewhat by the availability of recent academic work online. However, the kinds of text search systems in popular use today are poor at handling vocabulary mismatch, so a researcher must know the words used in relevant documents in order to find them. This makes serendipitous results unlikely. We approach the problem of literature search by considering an unpublished manuscript as a query to a search system. With this approach, the entire text content of the paper can be used in the search process. We use the text of previous literature as well as the citation graph that connects it to find relevant related material. We evaluate our technique with manual and automatic evaluation methods, and find an order of magnitude improvement in mean average precision as compared to a text similarity baseline. 1
Matching resumes and jobs based on relevance models.

A comparison of sentence retrieval techniques.

UMass at TREC 2007 Blog Distillation Task.

Identifying and improving retrieval for procedural questions.

Linear feature%2Dbased models for information retrieval.
Abstract. There have been a number of linear, feature-based models proposed by the information retrieval community recently. Although each model is presented differently, they all share a common underlying framework. In this paper we explore and discuss the theoretical issues of this framework, including a novel look at the parameter space. We then detail supervised training algorithms that directly maximize the evaluation metric under consideration, such as mean average precision. We present results that show training models in this way can lead to significantly better test set performance compared to other training methods that do not directly maximize the metric. Finally, we show that linear feature-based models can consistently and significantly outperform current state of the art retrieval models with the correct choice of features.
Beyond Bags of Words Modeling Implicit User Preferences in Information Retrieval.
This paper reports on recent work in the field of information  retrieval that attempts to go beyond the  overly simplified approach of representing documents  and queries as bags of words. Simple models make it  difficult to accurately model a user&#039;s information need.
Query Structuring with Two%2DStage Term Dependence in the Japanese Language.

Improving novelty detection for general topics using sentence level information patterns.
The detection of new information in a document stream is an important component of many potential applications. In this work, a new novelty detection approach based on the identification of sentence level information patterns is proposed. First, the information- pattern concept for novelty detection is presented with the emphasis on new information patterns for general topics (queries) that cannot be simply turned into specific questions whose answers are specific named entities (NEs). Then we elaborate a thorough analysis of sentence level information patterns on data from the TREC novelty tracks, including sentence lengths, named entities, sentence level opinion patterns. This analysis provides guidelines in applying those patterns in novelty detection particularly for the general topics. Finally, a unified pattern-based approach is presented to novelty detection for both general and specific topics. The new method for dealing with general topics will be the focus. Experimental results show that the proposed approach significantly improves the performance of novelty detection for general topics as well as the overall performance for all topics from the 2002-2004 TREC novelty tracks.
Ranking robustness a novel framework to predict query performance.

Boosting relevance model performance with query term dependence.

LDA%2Dbased document models for ad%2Dhoc retrieval.
Abstract—Semantic smoothing, which incorporates synonym and sense information into the language models, is effective and potentially significant to improve retrieval performance. Previously implemented semantic smoothing models such as the translation model have shown good experimental results. However, these models are unable to incorporate contextual information. To overcome this limitation, we propose a novel context-sensitive semantic smoothing method that decomposes a document into a set of weighted context-sensitive topic signatures and then maps those topic signatures into query terms. The language model with such a contextsensitive semantic smoothing is referred to as the topic signature language model. In detail, we implement two types of topic signatures, depending on whether ontology exists in the application domain. One is the ontology-based concept and the other is the multiword phrase. The mapping probabilities from each topic signature to individual terms are estimated through the EM algorithm. Document models based on topic signature mapping are then derived. The new smoothing method is evaluated on theTREC 2004/ 2005 Genomics Track with ontology-based concepts, as well as the TREC Ad Hoc Track (Disks 1, 2, and 3) with multiword phrases. Both experiments show significant improvements over the two-stage language model, as well as the language model with contextinsensitive semantic smoothing. Index Terms—Information retrieval, language model, semantic smoothing, topic signature, concept, multiword phrase. 1
Representing documents with named entities for story link detection %28SLD%29.

Hierarchical Language Models for Expert Finding in Enterprise Corpora.
Enterprise corpora contain evidence of what employees work on and therefore can be used to automatically find experts on a given topic. We present a general approach for representing the knowledge of a potential expert as a mixture of language models from associated documents. First we retrieve documents given the expert’s name using a generative probabilistic technique and weight the retrieved documents according to expert-specific posterior distribution. Then we model the expert indirectly through the set of associated documents, which allows us to exploit their underlying structure and complex language features. Experiments show that our method has excellent performance on TREC 2005 expert search task and that it effectively collects and combines evidence for expertise in a heterogeneous collection. 1
LDA%2Dbased document models for ad%2Dhoc retrieval.
Abstract—Semantic smoothing, which incorporates synonym and sense information into the language models, is effective and potentially significant to improve retrieval performance. Previously implemented semantic smoothing models such as the translation model have shown good experimental results. However, these models are unable to incorporate contextual information. To overcome this limitation, we propose a novel context-sensitive semantic smoothing method that decomposes a document into a set of weighted context-sensitive topic signatures and then maps those topic signatures into query terms. The language model with such a contextsensitive semantic smoothing is referred to as the topic signature language model. In detail, we implement two types of topic signatures, depending on whether ontology exists in the application domain. One is the ontology-based concept and the other is the multiword phrase. The mapping probabilities from each topic signature to individual terms are estimated through the EM algorithm. Document models based on topic signature mapping are then derived. The new smoothing method is evaluated on theTREC 2004/ 2005 Genomics Track with ontology-based concepts, as well as the TREC Ad Hoc Track (Disks 1, 2, and 3) with multiword phrases. Both experiments show significant improvements over the two-stage language model, as well as the language model with contextinsensitive semantic smoothing. Index Terms—Information retrieval, language model, semantic smoothing, topic signature, concept, multiword phrase. 1
A framework to predict the quality of answers with non%2Dtextual features.
New types of document collections are being developed by various web services. The service providers keep track of non-textual features such as click counts. In this paper, we present a framework to use non-textual features to predict the quality of documents. We also show our quality measure can be successfully incorporated into the language modeling-based retrieval model. We test our approach on a collection of question and answer pairs gathered from a community based question answering service where people ask and answer questions. Experimental results using our quality measure show a significant improvement over our baseline.
Representing clusters for retrieval.
In cluster-based retrieval (CBR), documents are grouped into clusters which are then retrieved directly in their entirety to the query or used to smooth document language models [1]. The clusters are typically represented as simple concatenation of their
Indri TREC Notebook 2006 Lessons Learned From Three Terabyte Tracks.
this submission we take a classification approach on document pairs to using relevance feedback information. We explore using textual and non-textual document-pair features to classify unjudged documents as relevant or non-relevant, and use this prediction to re-rank a baseline document retrieval. These features include co-citation measures, URL similarities, as well as features often used in machine learning systems for document ranking such as the difference in scores assigned by the baseline retrieval system. 1.
UMass at TREC 2006 Enterprise Track.

Table extraction for answer retrieval.

Precision prediction based on ranked list coherence.

Document quality models for web ad hoc retrieval.

Finding experts in community%2Dbased question%2Danswering services.

Document quality models for web ad hoc retrieval.

Similarity measures for tracking information flow.
Text similarity spans a spectrum, with broad topical similarity near one extreme and document identity at the other. Intermediate levels of similarity -- resulting from summarization, paraphrasing, copying, and stronger forms of topical relevance -- are useful for applications such as information flow analysis and question-answering tasks. In this paper, we explore mechanisms for measuring such intermediate kinds of similarity, focusing on the task of identifying where a particular piece of information originated. We consider both sentence-to-sentence and document-to-document comparison, and have incorporated these algorithms into RECAP, a prototype information flow analysis tool. Our experimental results with RECAP indicate that new mechanisms such as those we propose are likely to be more appropriate than existing methods for identifying the intermediate forms of similarity.
Novelty detection based on sentence level patterns.
The detection of new information in a document stream is an important component of many potential applications. In this paper, a new novelty detection approach based on the identification of sentence level patterns is proposed. Given a user’s information need, some patterns in sentences such as combinations of query words, named entities and phrases, may contain more important and relevant information than single words. Therefore, the proposed novelty detection approach focuses on the identification of previously unseen query-related patterns in sentences. Specifically, a query is preprocessed and represented with patterns that include both query words and required answer types. These patterns are used to retrieve sentences, which are then determined to be novel if it is likely that a new answer is present. An analysis of patterns in sentences was performed with data from the TREC 2002 novelty track and experiments on novelty detection were carried out on data from the TREC 2003 and 2004 novelty tracks. The experimental results show that the proposed pattern-based approach significantly outperforms all three baselines in terms of precision at top ranks.
Finding similar questions in large question and answer archives.
There has recently been a significant increase in the number of community-based question and answer services on the Web where people answer other peoples ’ questions. These services rapidly build up large archives of questions and answers, and these archives are a valuable linguistic resource. One of the major tasks in a question and answer service is to find questions in the archive that a semantically similar to a user’s question. This enables high quality answers from the archive to be retrieved and removes the time lag associated with a community-based system. In this paper, we discuss methods for question retrieval that are based on using the similarity between answers in the archive to estimate probabilities for a translation-based retrieval model. We show that with this model it is possible to find semantically similar questions with relatively little word overlap.
A Translation Model for Sentence Retrieval.
In this work we propose a translation model for monolingual sentence retrieval. We propose four methods for constructing a parallel corpus. Of the four methods proposed, a lexicon learned from a bilingual Arabic-English corpus aligned at the sentence level performs best, significantly improving results over the query likelihood baseline. Further, we demonstrate that smoothing from the local context of the sentence improves retrieval over the query likelihood baseline. 1
Optimization strategies for complex queries.
this paper, we propose a set of strategies for processing this new class of path queries based on varying design decisions along two dimensions. The two dimensions are (1) the spatial overlap operations between areas and links can be preprocessed or intermixed with the path search process and (2) areas satisfying the query constraint can be prefiltered or dynamically selected during path search. We then propose efficient techniques for implementing the resulting four integrated query processing strategies, utilizing technologies such as spatial joins, R-tree for spatial overlap search, and spatial clustering for efficient path search. In this paper, we conduct extensive experiments for our proposed strategies and evaluate their performance based on I/O efficiency. Our experimental studies produce several interesting observations. For example, we find that (1) preprocessing spatial queries is advantageous for smaller buffer sizes, and a low area selectivity factor even when the preprocessing cost is included, and (2) intermixing path search with spatial queries works better with larger buffer sizes, and a high area selectivity factor.
A Markov random field model for term dependencies.
This paper develops a general, formal framework for modeling term dependencies via Markov random fields. The model allows for arbitrary text features to be incorporated as evidence. In particular, we make use of features based on occurrences of single terms, ordered phrases, and unordered phrases. We explore full independence, sequential dependence, and full dependence variants of the model. A novel approach is developed to train the model that directly maximizes the mean average precision rather than maximizing the likelihood of the training data. Ad hoc retrieval experiments are presented on several newswire and web collections, including the GOV2 collection used at the TREC 2004 Terabyte Track. The results show significant improvements are possible by modeling dependencies, especially on the larger web collections.
Finding semantically similar questions based on their answers.
A large number of question and answer pairs can be collected from question and answer boards and FAQ pages on the Web. This paper proposes an automatic method of finding the questions that have the same meaning. The method can detect semantically similar questions that have little word overlap because it calculates question-question similarities by using the corresponding answers as well as the questions. We develop two different similarity measures based on language modeling and compare them with the traditional similarity measures. Experimental results show that semantically similar questions pairs can be effectively found with the proposed similarity measures.
The recap system for identifying information flow.

Indri at TREC 2005 Terabyte Track.
This work details the experiments carried out using the Indri search engine during the TREC 2005 Terabyte Track. Results are presented for each of the three tasks, including e#ciency, ad hoc, and named page finding. Our e#ciency runs focused on query optimization techniques, our ad hoc runs look at the importance of term proximity and document quality, and our named-page finding runs investigate the use of document priors and document structure.
Indri at TREC 2005 Terabyte Track.
This work details the experiments carried out using the Indri search engine during the TREC 2005 Terabyte Track. Results are presented for each of the three tasks, including e#ciency, ad hoc, and named page finding. Our e#ciency runs focused on query optimization techniques, our ad hoc runs look at the importance of term proximity and document quality, and our named-page finding runs investigate the use of document priors and document structure.
UMass Robust 2005 Using Mixtures of Relevance Models for Query Expansion.
This paper describes the UMass TREC 2005 Robust Track experiments. We focus on approaches that use term proximity and pseudo-relevance feedback using external collections. Our results indicate both approaches are highly e#ective.
The Lowell database research self%2Dassessment.

Analysis of Statistical Question Classification for Fact%2DBased Questions.
Question classification systems play an important role in question answering systems and can be used in a wide range of other domains. The goal of question classification is to accurately assign labels to questions based on expected answer type. Most approaches in the past have relied on matching questions against hand-crafted rules. However, rules require laborious e#ort to create and often su#er from being too specific. Statistical question classification methods overcome these issues by employing machine learning techniques. We empirically show that a statistical approach is robust and achieves good performance on three diverse data sets with little or no hand tuning. Furthermore, we examine the role di#erent syntactic and semantic features have on performance. We find that semantic features tend to increase performance more than purely syntactic features. Finally, we analyze common causes of misclassification error and provide insight into ways they may be overcome.
Computer%2DAssisted Information Retrieval %28Recherche d%27Information et ses Applications%29 %2D RIAO 2004

A Multi%2DAgent Approach for Peer%2Dto%2DPeer Based Information Retrieval System.

A framework for selective query expansion.
Query expansion is a well-known technique that has been shown to improve average retrieval performance. This technique has not been used in many operational systems because of the fact that it can greatly degrade the performance of some individual queries. We show how comparison between language models of the unexpanded and expanded retrieval results can be used to predict when the expanded retrieval has strayed from the original sense of the query. In these cases, the unexpanded results are used while the expanded results are used in the remaining cases (where such straying is not detected). We evaluate this method on a wide variety of TREC collections.
A Language Modeling Approach to Metadata for Cross%2DDatabase Linkage and Search.

Question Answering Performance on Table Data.
Question answering (QA) on table data is a challenging information retrieval task. This paper describes a QA system for tables created with both machine learning and heuristic table extraction methods. Errors were analyzed in order to improve the system using government statistical data. We also apply these improvements on another type of table data set and show the experimental results.
Cluster%2Dbased retrieval using language models.
Previous research on cluster-based retrieval has been inconclusive as to whether it does bring improved retrieval effectiveness over document-based retrieval. Recent developments in the language modeling approach to IR have motivated us to re-examine this problem within this new retrieval framework. We propose two new models for cluster-based retrieval and evaluate them on several TREC collections. We show that cluster-based retrieval can perform consistently across collections of realistic size, and significant improvements over document-based retrieval can be obtained in a fully automatic manner and without relevance information provided by human.
Formal multiple%2Dbernoulli models for language modeling.
In language modeling, it is nearly always assumed that documents are generated by sampling from a multinomial distribution. Many formal methods of estimation exist for the multinomial case. In this paper, we reexamine language models based on a multiple-Bernoulli distribution. This assumption has been explored in the past, but has never been formalized. Here, we present how smoothed language models can be estimated using this assumption from a formal, Bayesian standpoint.
Evaluating high accuracy retrieval techniques.
ABSTRACT Although information retrieval research has always been concernedwith improving the effectiveness of search, in some applications, such as information analysis, a more specific requirement exists forhigh accuracy retrieval. This means that achieving high precision in the top document ranks is paramount. In this paper we presentwork aimed at achieving high accuracy in ad-hoc document retrieval by incorporating approaches from question answering (QA).We focus on getting the first relevant result as high as possible in the ranked list and argue that traditional precision and recall are notappropriate measures for evaluating this task. We instead use the mean reciprocal rank (MRR) of the first relevant result. We eval-uate three different methods for modifying queries to achieve high accuracy. The experiments done on TREC data provide support forthe approach of using MRR and incorporating QA techniques for getting high accuracy in ad-hoc retrieval task. Categories and Subject Descriptors H.3.4 [Information Storage and Retrieval]: Systems and Soft-ware--Performance evaluation (efficiency and effectiveness) ; H.3.3 [Information Storage and Retrieval]: Information Search andRetrieval--Query formulation
Answer models for question answering passage retrieval.
Answer patterns have been shown to improve the performance of open-domain factoid QA systems. Their use, however, requires either constructing the patterns manually or developing algorithms for learning them automatically. We present here a simpler approach that extends the techniques of language modeling to create answer models. These are language models trained on the correct answers to training questions. We show how they fit naturally into a probabilistic model for answer passage retrieval and demonstrate their effectiveness on the TREC 2002 QA Corpus.
Formal multiple%2Dbernoulli models for language modeling.
In language modeling, it is nearly always assumed that documents are generated by sampling from a multinomial distribution. Many formal methods of estimation exist for the multinomial case. In this paper, we reexamine language models based on a multiple-Bernoulli distribution. This assumption has been explored in the past, but has never been formalized. Here, we present how smoothed language models can be estimated using this assumption from a formal, Bayesian standpoint.
Automatic recognition of reading levels from user queries.
INTRODUCTION  In general, interactions with Web search engines could be characterized as &#034;one size fits all&#034;. There is no representation of user preferences, search context, or the task context. Because of this obvious limitation of current search technology, personalization and context have been identified as major research challenges in a number of workshops. This work describes an ongoing effort toward automatically inferring knowledge about users and context from user queries. We report experiments that focus on the recognition of one context feature, namely reading level, and show that queries from users of different level groups can be effectively separated.  Scientific readablility [4] has been studied for decades. Many readability indices have been developed to determine the reading levels of texts. Popular ones include Flesch-Kincaid [3], SMOG [5], and FOG [2] tests. A statistical approach has been proposed recently in [6]. A common characteristic among all these methods is th
Indri at TREC 2004 Terabyte Track.
This paper provides an overview of experiments carried out at the TREC 2004 Terabyte Track using the Indri search engine. Indri is an efficient, effective distributed search engine. Like INQUERY, it is based on the inference network framework and supports structured queries, but unlike INQUERY, it uses language modeling probabilities within the network which allows for added flexibility. We describe our approaches to the Terabyte Track, all of which involved automatically constructing structured queries from the title portions of the TREC topics. Our methods use term proximity information and HTML document structure. In addition, a number of optimization procedures for efficient query processing are explained.
UMass at TREC 2004 Novelty and HARD.
For the TREC 2004 Novelty track, UMass participated in all four tasks. Although finding relevant sentences was harder this year than last, we continue to show marked improvements over the baseline of calling all sentences relevant, with a variant of tfidf being the most successful approach. We achieve 5–9 % improvements over the baseline in locating novel sentences, primarily by looking at the similarity of a sentence to earlier sentences and focusing on named entities. For the High Accuracy Retrieval from Documents (HARD) track, we investigated the use of clarification forms, fixed- and variable-length passage retrieval, and the use of metadata. Clarification form results indicate that passage level feedback can provide improvements comparable to user supplied related-text for document evaluation and outperforms related-text for passage evaluation. Document retrieval methods without a query expansion component show the most gains from related-text. We also found that displaying the top passages for feedback outperformed displaying centroid passages. Named entity feedback resulted in mixed performance. Our primary findings for passage retrieval are that document retrieval methods performed better than passage retrieval methods on the passage evaluation metric of binary preference at 12,000 characters, and that clarification forms improved passage retrieval for every retrieval method explored. We found no benefit to using variable-length passages over fixed-length passages for this corpus. Our use of geography and genre metadata resulted in no significant changes in retrieval performance.
Combining the language model and inference network approaches to retrieval.
This paper describes the development of a structured document collection containing user-generated text and numerical metadata for exploring the exploitation of metadata in information retrieval (IR). The collection consists of more than 61,000 documents extracted from YouTube video pages on basketball in general and NBA (National Basketball Association) in particular, together with a set of 40 topics and their relevance judgements. In addition, a collection of nearly 250,000 user profiles related to the NBA collection is available. Several baseline IR experiments report the effect of using video-associated metadata on retrieval effectiveness. The results surprisingly show that searching the videos titles only performs significantly better than searching additional metadata text fields of the videos such as the tags or the description. 1.
Thirteenth International Work Shop on Research Issues in Data Engineering Multi%2Dlingual Information Management

Time%2Dbased language models.
We explore the relationship between time and relevance using TREC ad-hoc queries. A type of query is identified that favors very recent documents. We propose a time-based language model approach to retrieval for these queries. We show how time can be incorporated into both query-likelihood models and relevance models. We carried out experiments to compare time-based language models to heuristic techniques for incorporating document recency in the ranking. Our results show that time-based models perform as well as or better than the best of the heuristic techniques.
Relevant query feedback in statistical language modeling.
In traditional relevance feedback, researchers have explored relevant document feedback, wherein, the query representation is updated based on a set of relevant documents returned by the user. In this work, we investigate relevant query feedback, in which we update a document&#039;s representation based on a set of relevant queries. We propose four statistical models to incorporate relevant query feedback.
Table Extraction Using Conditional Random Fields.
The ability to find tables and extract information from them is a necessary component of data mining, question answering, and other information retrieval tasks. Documents often contain tables in order to communicate densely packed, multi-dimensional information. Tables do this by employing layout patterns to efficiently indicate fields and records in two-dimensional form.
Language Models for Information Retrieval.
We systematically investigate a new approach to estimating the parameters of language models for information retrieval, called parsimonious language models. Parsimonious language models explicitly address the relation between levels of language models that are typically used for smoothing. As such, they need fewer (non-zero) parameters to describe the data. We apply parsimonious models at three stages of the retrieval process: 1) at indexing time; 2) at search time; 3) at feedback time. Experimental results show that we are able to build models that are significantly smaller than standard models, but that still perform at least as well as the standard approaches.
Salton Award Lecture %2D Information retrieval and computer science an evolving relationship.

Table extraction using conditional random fields.
The ability to find tables and extract information from them is a necessary component of data mining, question answering, and other information retrieval tasks. Documents often contain tables in order to communicate densely packed, multi-dimensional information. Tables do this by employing layout patterns to efficiently indicate fields and records in two-dimensional form.
Generating hierarchical summaries for web searches.
Hierarchies provide a means of organizing, summarizing and accessing information. We describe a method for automatically generating hierarchies from small collections of text, and then apply this technique to summarizing the documents retrieved by a search engine. We show that these hierarchies provide better access to the documents than a simple ranked list and that the terms in the hierarchy are better summaries of the documents than the top TF.IDF weighted terms. In addition, we discuss the formal framework of the technique and how the technique has been used with news databases and TREC collections.
The Lowell Database Research Self Assessment

Challenges in information retrieval and language modeling report of a workshop held at the center for intelligent information retrieval

Passage retrieval based on language models.
Previous research has shown that passage-level evidence can bring added benefits to document retrieval when documents are long or span different subject areas. Recent developments in language modeling approach to IR provided a new effective alternative to traditional retrieval models. These two streams of  research motivate us to examine the use of passages in a language model framework. This paper reports on experiments using passages in a simple language model and a relevance model, and compares the results with document-based retrieval. Results from the INQUERY search engine, which is not based on a language modeling approach, are also given for comparison. Test data include two heterogeneous and one homogeneous document collections. Our experiments show that passage retrieval is feasible in the language modeling context, and more importantly, it can provide more reliable performance than retrieval based on full documents.
Features of documents relevant to task%2D and fact%2Doriented questions.
We describe results from an ongoing project that considers question  types and document features and their relationship to retrieval techniques. We examine eight document features from the top 25  documents retrieved from 74 questions and find that lists and  FAQs occur in more documents judged relevant to task-oriented  questions than those judged relevant to fact-oriented questions.
QuASM a system for question answering using semi%2Dstructured data.
This paper describes a system for question answering using semistructured metadata, QuASM (pronounced &#034;chasm&#034;). Question answering systems aim to improve search performance by providing users with specific answers, rather than having users scan retrieved documents for these answers. Our goal is to answer factual questions by exploiting the structure inherent in documents found on the World Wide Web (WWW). Based on this structure, documents are indexed into smaller units and associated with metadata. Transforming table cells into smaller units associated with metadata is an important part of this task. In addition, we report on work to improve question classification using language models. The domain used to develop this system is documents retrieved from a crawl of www.fedstats.gov.
Cross%2Dlingual relevance models.
We propose a formal model of Cross-Language Information Retrieval that does not rely on either query translation or document translation. Our approach leverages recent advances in language modeling to directly estimate an accurate topic model in the target language, starting with a query in the source language. The model integrates popular techniques of disambiguation and query expansion in a unified formal framework. We describe how the topic model can be estimated with either a parallel corpus or a dictionary. We test the framework by constructing Chinese topic models from English queries and using them in the CLIR task of TREC9. The model achieves performance around 95% of the strong mono-lingual baseline in terms of average precision. In initial precision, our model outperforms the monolingual baseline by 20%. The main contribution of this work is the unified formal model which integrates techniques that are essential for e#ective Cross-Language Retrieval.
Predicting query performance.
We develop a method for predicting query performance by computing the relative entropy between a query language model and the corresponding collection language model. The resulting clarity score measures the coherence of the language usage in documents whose models are likely to generate the query. We suggest that clarity scores measure the ambiguity of a query with respect to a collection of documents and show that they correlate positively with average precision in a variety of TREC test sets. Thus, the clarity score may be used to identify ine#ective queries, on average, without relevance information. We develop an algorithm for automatically setting the clarity score threshold between predicted poorly-performing queries and acceptable queries and validate it using TREC data. In particular, we compare the automatic thresholds to optimum thresholds and also check how frequently results as good are achieved in sampling experiments that randomly assign queries to the two classes.
Task orientation in question answering.
INTRODUCTION  Information retrieval techniques treat questions democratically: all questions, regardless of their grammar or orientation, are processed by the same rules and manipulated in similar ways. In about 12% of the questions in our data users ask about a process. Treating a request for how to do something as a request for information about a topic disregards an important subtlety in the question. We use the knowledge that a question asks about a process, rather than a fact, to develop categories of information need implied by the questions to aid in retrieval. We focus on questions because by their nature questions carry more information.  A well-known technique for improving retrieval is query expansion. A variety of query expansion and disambiguation techniques have been studied [1] [2] [3] [5], and although none distinguish between fact-orientation and taskorientation, several do distinguish categories of information need not explicit in the queries.  Understanding that the 
A survey in indexing and searching XML documents.

SIGIR 2001 Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval

Relevance Feedback and Personalization A Language Modeling Perspective.
Many approaches to personalization involve learning short-term and long-term user models. The user models provide context for queries and other interactions with the information system. In this paper, we discuss how language models can be used to represent context and support context-based techniques such as relevance feedback and query disambiguation.  1. Overview  From some perspectives, personalization has been studied in information retrieval for some time. If the goal of personalization is to improve the effectiveness of information access by adapting to individual users&#039; needs, then techniques such as relevance feedback and filtering would certainly be considered to support personalization. There has also been considerable research done, mostly in the 1980s, on user modeling for information retrieval. This research had essentially the same goal as current research on personalization, which is to build a model of a user&#039;s interests and preferences over time. Filtering systems, too...
Relevance%2DBased Language Models.
It has long been recognized that the primary obstacle to effective performance of classical models is the need to estimate a relevance model with no training data. We propose a novel technique for estimating such models using the query alone. We demonstrate that our technique can produce highly accurate relevance models. Our experiments show relevance models outperforming baseline language modeling systems on TREC retrieval. The main contribution of this work is an effective formal method for estimating a relevance model with no training data.  1 
Finding Topic Words for Hierarchical Summarization.

Probabilistic techniques for phrase extraction.

Workshop on language modeling and information retrieval.
We incorporate stemming into the language modeling framework. The work is suggested by the notion that stemming increases the numbers of word occurrences used to estimate the probability of a word (by including the members of its stem class). As such, stemming can be viewed as a type of smoothing of probability estimates. We show that such a view of stemming leads to a simple incorporation of ideas from corpus-based stemming. We also present two generative models of stemming. The first generates terms and then variant stems. The second generates stem classes and then a member. All models are evaluated empirically, though there is little di#erence between the various forms of stemming.
Combining Approaches to IR %28invited talk%29.


Information Retrieval Based on Statistical Language Models.
This paper follows a formal approach to information retrieval based on statistical language models. By introducing some simple reformulations of the basic language modeling approach we introduce the notion of importance of a query term. The importance of a query term is an unknown parameter that explicitly models which of the query terms are generated from the relevant documents (the important terms), and which are not (the unimportant terms). The new language modeling approach is shown to explain a number of practical facts of today&#039;s information retrieval systems that are not very well explained by the current state of information retrieval theory, including stop words, mandatory terms, coordination level ranking and retrieval using phrases.
An exploratory analysis of phrases in text retrieval.
Phrases are used in both commercial and experimental search engines. Despite the large amount  of work in the area results remain mixed. It is still not clear whether phrases can be used to improve  retrieval effectiveness. In this paper, we examine phrases and their properties independently of any  specific retrieval approach. We explore phrase usage in text corpora and relevance patterns related to  phrase usage. The result is not only a better understanding of phrases, but a better method by which  phrases and phrase techniques may be evaluated. With this method we can directly determine the  value of various phrase formulations for information retrieval.  1 Introduction  Phrases in information retrieval have been used to aid effectiveness. They have gone by many names: multiword features, phrase terms, nominal compounds and so on. Despite their many implementations, retrieval experiments with phrases are often inconclusive. In some experiments phrases yield small improvements in re...

INQUERY and TREC%2D9.

Improving the effectiveness of information retrieval with local context analysis.
Techniques for automatic query expansion have been extensively studied in information  retrieval research as a means of addressing the word mismatch between queries and  documents. These techniques can categorized as either global or local. While global  techniques rely on analysis of a whole collection to discover word relationships, local  techniques emphasize analysis of the top ranked documents retrieved for a query. Both  types of techniques have advantages and limitations. In this paper we propose a new  technique, called local context analysis, which combines the advantages of a global technique  called Phrasefinder and a local technique known as local feedback. Experiments on  a number of collections, both English and non-English, show that local context analysis  offers more effective and consistent retrieval results.
A General Language Model for Information Retrieval.
Statistical language modeling has been successfully used for speech recognition, part-of-speech tagging, and syntactic parsing. Recently, it has also been applied to information retrieval. According to this new paradigm, each document is viewed as a language sample, and a query as a generation process. The retrieved documents are ranked based on the probabilities of producing a query from the corresponding language models of these documents. In this paper, we will present a new language model for information retrieval, which is based on a range of data smoothing techniques, including the Good-Turing estimate, curve-fitting functions, and model combinations. Our model is conceptually simple and intuitive, and can be easily extended to incorporate probabilities of phrases such as word pairs and word triples. The experiments with the Wall Street Journal and TREC4 data sets showed that the performance of our model is comparable to that of INQUERY and better than that of another language model for information retrieval. In particular, word pairs are shown to be useful in improving the retrieval performance.
Deriving Concept Hierarchies from Text.

Cluster%2DBased Language Models for Distributed Retrieval.
Effective retrieval in a distributed environment is an important but difficult problem. Lack of effectiveness appears to have three causes. First, collection selection based on word histograms is not appropriate for heterogeneous collections. Second, relevant documents are scattered over many collections and searching a few collections misses many relevant documents. Third, most existing collection selection metrics lack sound theoretical justifications and hence may not be well tuned to the problem. We propose a new approach to distributed retrieval based on document clustering and language modeling. Document clustering is used to organize collections around topics. Language modeling is used to properly represent topics and effectively select the right topics for a query. Based on these ideas, three methods are proposed to suit different environments. We show that all three methods improve effectiveness of distributed retrieval.
A General Language Model for Information Retrieval %28poster abstract%29.

PIC matrices a computationally tractable class of probabilistic query operators.
this paper, we present a class of query operators which, for reasons to be explained, we have called pic operators. The pic operators comprise a subclass of the general class of query operators that are expressible within the inference network framework. The subclass has been chosen to satisfy two important design criteria: 1) the operators belonging to the subclass are computationally tractable, and 2) they are intuitively plausible candidates for the modeling of query operators. A wide variety of query operators can be defined as pic operators. The work reported has been principally motivated by the search for more effective Boolean operators. Other approaches to the modeling of Boolean query operators have been tried. As with the inference network, the goal has been to generalize the classical, strict Propositional Logic interpretation of the query operators. The objective of this generalization is twofold. On the one hand, to allow for graduated inputs to the Boolean operators so that the representation of documents in terms of vectors of Boolean characteristics can be extended to vectors of feature weights. Second, to generate real valued operator output so that dichotomous relevance judgments can be replaced by document ranking in response to user&#039;s queries. A particularly notable success in this pursuit was reported by Salton, Fox and Wu [Salton et al. 1983; Salton et al. 1983]. Grounded in the geometric metaphor of the vector space model, they defined a general class of &#034;pnorm&#034; operators which extend the traditional operators in a natural way. The experimental results achieved were quite positive. Recent experimentation has shown that system performance is improved by replacing the inference network Boolean operator calculation used in the inquery retrieval syst...
A Language Modeling Approach to Information Retrieval.
Models of document indexing and document retrieval have been extensively studied. The integration of these two classes of models has been the goal of several researchers but it is a very difficult problem. We argue that much of the reason for this is the lack of an adequate indexing model. This suggests that perhaps a better indexing model would help solve the problem. However, we feel that making unwarranted parametric assumptions will not lead to better retrieval performance. Furthermore, making prior assumptions about the similarity of documents is not warranted either. Instead, we propose an approach to retrieval based on probabilistic language modeling. We estimate models for each document individually. Our approach to modeling is non-parametric and integrates document indexing and document retrieval into a single model. One advantage of our approach is that collection statistics which are used heuristically in many other retrieval models are an integral part of our model. We have...
Resolving Ambiguity for Cross%2DLanguage Retrieval.
One of the main hurdles to improved CLIR effectiveness is resolving ambiguity associated with translation. Availability of resources is also a problem. First we present a technique based on co-occurrence statistics from unlinked corpora which can be used to reduce the ambiguity associated with phrasal and term translation. We then combine this method with other techniques for reducing ambiguity and achieve more than 90% monolingual effectiveness. Finally, we compare the co-occurrence method with parallel corpus and machine translation techniques and show that good retrieval effectiveness can be achieved without complex resources.  1 
Corpus%2DBased Stemming Using Cooccurrence of Word Cariants.

Sorting Out Searching A User%2DInterface Framework for Text Searches.
: Current user interfaces for textual database searching leave much to be desired: individually, they are often confusing, and as a group, they are seriously inconsistent. We propose a four-phase framework for user-interface design. The framework provides common structure and terminology for searching while preserving the distinct features of individual collections and search mechanisms. This paper appeared in CACM 41, 4 (April 1998), pp. 95--98. An earlier but more detailed version appeared as Tech Report IR-107 in this series, and online in D-Lib Magazine:  http://www.dlib.org/dlib/january97/01contents.html This material is based on work supported in part by the National Science Foundation, Library of Congress, and Department of Commerce under cooperative agreement number EEC-9209623,by NRaD Contract Number N66001-94-D-6054, by NASA contract NAG-528-95, and by NSF contract IRI-96-15534. Any opinions, findings, conclusions and/or recommendations expressed herein are those of the authors and may not reflect those of the sponsors.  2  
TINTIN A System for Retrieval in Text Tables.
Tables form an important kind of data element in text retrieval. Often, the gist of an entire  news article or other exposition can be concisely captured in tabular form. In this paper, we  examine the utility of exploiting information other than the key words in a digital document  to provide the users with more flexible and powerful query capabilities. More specifically,  we exploit the structural information in a document to identify tables and their component  fields and let the users query based on these fields. Our empirical results have demonstrated  that heuristic method based table extraction and component tagging can be performed effectively  and efficiently. Moreover, our experiments in retrieval using the TINTIN system have  strongly indicated that such structural decomposition can facilitate better representation of  user&#039;s information needs and hence more effective retrieval of tables.  1 Introduction  Much of the current literature in information retrieval deals with wor...
Text Segmentation by Topic.
. We investigate the problem of text segmentation by topic. Applications for this task include topic tracking of broadcast speech data and topic identification in full-text databases. Researchers have tackled similar problems before but with different goals. This study focuses on data with relatively small segment sizes and for which within-segment sentences have relatively few words in common making the problem challenging. We present a method for segmentation which makes use of a query expansion technique to find common features for the topic segments. Experiments with the technique show that it can be effective. 1 Introduction  There has recently been increased interest in tracking topics in information &#034;feeds.&#034; A feed is a continuous stream of text produced, for example, by speech recognition of broadcast news. The text in such an environment contains no mark-up to indicate topic boundaries and may not even delineate sentences. The text also tends to be terse and words are not repe...
Probabilistic Retrieval of OCR Degraded Text Using N%2DGrams.
. The retrieval of OCR degraded text using n-gram formulations within a probabilistic retrieval system is examined in this paper. Direct retrieval of documents using n-gram databases of 2 and 3-grams or 2, 3, 4 and 5-grams resulted in improved retrieval performance over standard (word based) queries on the same data when a level of 10 percent degradation or worse was achieved. A second method of using n-grams to identify appropriate matching and near matching terms for query expansion which also performed better than using standard queries is also described. This method was less effective than direct n-gram query formulations but can likely be improved with alternative query component weighting schemes and measures of term similarity. Finally, a web based retrieval application using n-gram retrieval of OCR text and display, with query term highlighting, of the source document image is described. 1 Introduction  A major problem with retrieval of OCR text from image data is the inevitabl...
Computationally Tractable Probabilistic Modeling of Boolean Operators.
The inference network model of information retrieval allows for a probabilistic interpretation of Boolean query operators. Prior work has shown, however, that these operators do not perform as well as the the pnorm operators developed in the context of the vector space model. The design of alternative operators in the inference network framework must contend with the issue of computational tractability. We define a flexible class of link matrices that are natural candidates for the implementation of Boolean operators and an O(n  2  ) algorithm for the computation of probabilities involving link matrices of this class. We present experimental results indicating that Boolean operators implemented in terms of link matrices from this class perform at least as well as pnorm operators. 1 Introduction  We present a class of operators for the modeling of Boolean queries that is:  ffl intuitively plausible,  ffl probabilistically motivated, and  ffl computationally tractable. For reasons to be ...
Phrasal Translation and Query Expansion Techniques for Cross%2Dlanguage Information Retrieval.
Dictionary methods for cross-language information retrieval give performance below that for mono-lingual retrieval. Failure to translate multi-term phrases has been shown to be one of the factors responsible for the errors associated with dictionary methods. First, we study the importance of phrasal translation for this approach. Second, we explore the role of phrases in query expansion via local context analysis and local feedback and show how they can be used to significantly reduce the error associated with automatic dictionary translation. 1 Introduction  The development of IR systems for languages other than English has focused on building mono-lingual systems. Increased availability of on-line text in languages other than English and increased multi-national collaboration have motivated research in cross-language information retrieval (CLIR) - the development of systems to perform retrieval across languages. There have been three main approaches to CLIR: translation via machine t...
INQUERY Does Battle With TREC%2D6.
 
Dictionary Methods for Cross%2DLingual Information Retrieval.
. Multi-lingual information retrieval (IR) has largely been limited to the development of systems for use with a specific foreign language. The explosion in the availability of electronic media in languages other than English makes the development of IR systems that can cross language boundaries increasingly important. In this paper, we present experiments that analyze the factors that affect dictionary based methods for cross-lingual retrieval and present methods that dramatically reduce the errors such an approach usually makes. 1 Introduction  In recent years, the amount of online information from the government, scientific, and business communities has risen dramatically. In response, much work has been done to develop systems that provide effective and efficient access to electronic media. However, the diversity of information sources and the explosive growth of the Internet worldwide are compelling evidence of the need for IR systems that cross language boundaries. Increased inte...
Indexing Handwriting Using Word Matching.
There are many historical manuscripts written in a single hand which it would be useful to index. Examples include the W. B. DuBois collection at the University of Massachusetts and the early Presidential libraries at the Library of Congress. The standard technique for indexing documents is to scan them in, convert them to machine readable form (ASCII) using Optical Character Recognition (OCR) and then index them using a text retrieval engine. However, OCR does not work well on handwriting. Here an alternative scheme is proposed for indexing such texts. Each page of the document is segmented into words. The images of the words are then matched against each other to create equivalence classes (each equivalence classes contains multiple instances of the same word). The user then provides ASCII equivalents for say the top 2000 equivalence classes. The current paper deals with the matching aspects of this process. Due to variations in even a single person&#039;s handwriting, it is expected that...
Applications of Multilingual Text Retrieval.
The recent enormous increase in the use of networked information access and on-line databases has led to more databases being available in languages other than English. The Center for Intelligent Information Retrieval (CIIR) at the University of Massachusetts is involved in a variety of industrial, government, and digital library applications which have a need for multilingual text retrieval. Most information retrieval research, however, has been evaluated using English databases and queries, and relatively little is known about how well advanced statistical techniques that incorporate ranking and term weighting perform in different languages. We describe our experience with a range of projects involving text retrieval in Spanish, Japanese and Chinese. The issues covered by these projects include document representation techniques such as morphology and segmentation, query formulation and expansion techniques, relevance feedback, and comparisons of retrieval effectiveness with English ...
Combining Classifiers in Text Categorization.
Three different types of classifiers were investigated in the context of a text categorization problem in the medical domain: the automatic assignment of ICD9 codes to dictated inpatient discharge summaries. K-nearest-neighbor, relevance feedback, and Bayesian independence classifers were applied individually and in combination. A combination of different classifiers produced better results than any single type of classifier. For this specific medical categorization problem, new query formulation and weighting methods used in the k-nearest-neighbor classifier improved performance.   1 Introduction  Past research in information retrieval has shown that one can improve retrieval effectiveness by using multiple representations in indexing and query formulation [27] [19] [3] [11] and by using multiple search strategies [5] [24] [7]. In this work, we investigate whether we can attain similar improvements in the domain of text categorization by combining different representations and classif...
Query Expansion Using Local and Global Document Analysis.
Automatic query expansion has long been suggested as a technique for dealing with the fundamental issue of word mismatch in information retrieval. A number of approaches to expansion have been studied and, more recently, attention has focused on techniques that analyze the corpus to discover word relationships (global techniques) and those that analyze documents retrieved by the initial query ( local feedback). In this paper, we compare the effectiveness of these approaches and show that, although global analysis has some advantages, local analysis is generally more effective. We also show that using global analysis techniques, such as word context and phrase structure, on the local set of documents produces results that are both more effective and more predictable than simple local feedback. 1 Introduction The problem of word mismatch is fundamental to information retrieval. Simply stated, it means that people often use different words to describe concepts in their queries than auth...

INQUERY at TREC%2D5.
this paper focuses on relevant differences to the previously published algorithms. 1 Ad-Hoc Experiments
Uncertainty in Information Retrieval Systems.
This paper surveys methods for representing and reasoning with imperfect information. It opens with an attempt to classify the different types of imperfection that may pervade data, and a discussion of the sources of such imperfections. The classification is then used as a framework for considering work that explicitly concerns the representation of imperfect information, and related work on how imperfect information may be used as a basis for reasoning. The work that is surveyed is drawn from both the field of databases and the field of artificial intelligence. Both of these areas have long been concerned with the problems caused by imperfect information, and this paper stresses the relationships between the approaches developed in each.

Providing Government Information on the Internet Experiences with THOMAS.
The THOMAS system is designed to make legislative information available to the general public over the Internet, and can be regarded as a prototype of a government digital library. As a joint project between the Library of Congress and the University of Massachusetts, THOMAS provides significant opportunities for studying the retrieval and interface techniques that are needed to support effective access to complex information in an Internet-based environment. Preliminary experience with THOMAS indicates that there is significant interest in this type of information, and that queries tend to be shorter than those studied in retrieval evaluations such as TREC. Techniques such as query processing, query expansion, and morphological processing all need to be incorporated and improved. KEYWORDS: Information Retrieval, Internet, Query Processing INTRODUCTION In mid-December 1994, the Library of Congress was requested by the newly-elected leadership of the House of Representatives to develo...
Machine Learning and Information Retrieval %28Abstract%29.

Searching Distributed Collections with Inference Networks.
The use of information retrieval systems in networked environments raises a new set of issues that have received little attention. These issues include ranking document collections for relevance to a query, selecting the best set of collections from a ranked list, and merging the document rankings that are returned from a set of collections. This paper describes methods of addressing each issue in the inference network model, discusses their implementation in the INQUERY system, and presents experimental results demonstrating their effectiveness.  
Integrating IR and RDBMS Using Cooperative Indexing.

Recent Experiments with INQUERY.
this paper focuses on relevant differences to the previously published algorithms. 1 Description of Ad-Hoc Experiments
NSF Center for Intelligent Information Retrieval.
Recent explosion in networked information resources has been attracting much attention to the problem of automated methods for gathering information in response to a query from a user. However, most of this literature deals with locating, gathering and selecting the best response to a query from among a multitude of responses from different sources. In this paper, we deal with a different model of response to a query. No single source of information may contain the complete response to a query and hence may necessitate piecing together mutually related partial responses from disparate and possibly heterogeneous sources. We present a system for cooperative retrieval and composition of a case in which sub-cases are distributed across different agents in a multi-agent system. From a Gestalt perspective, a good overall case may not be the one derived from the summation of best subcases. Each agent&#039;s partial view may result in local cases that are best matches based on the local view. However, these local cases when assembled may not result in the best overall case in terms of global measures. We propose a negotiation-driven case retrieval algorithm as an approach to dynamically resolving inconsistencies between different case pieces during the retrieval process.   The work reported here is supported in part by NSF Center for Intelligent Information Retrieval, ARPA contract N00014-92-J-1698, Office of Naval Research contract N00014-92-J-1450, the National Science Foundation contract CDA 8922572. The content of the information does not necessarily reflect the position or the policy of the Government and no official endorsement should be inferred.  1 
From Office Automation to Intelligent Workflow Systems.

Effective Text Retrieval Based on Combining Evidence from the Corpus and Users.

TREC and Tipster Experiments with Inquery.
INQUERY is a probablistic information retrieval system based upon a Bayesian inference network model. This paper describes recent improvements to the system as a result of participation in the TIPSTER project and the TREC-2 conference. Improvements include transforming forms-based specifications of information needs into complex structured queries, automatic query expansion, automatic recognition of features in documents, relevance feedback, and simulated document routing. Experiments with one and two gigabyte document collections are also described. To appear in Information Processing and Management.  1 Introduction  The effectiveness of an information retrieval (IR) system depends upon representation and matching. The system must represent the information need, it must represent the documents, and it must determine how well the information need matches each document. Our approach has been to use improved representations of document text and queries in the framework of the inference ...
Combining Automatic and Manual Index Representations in Probabilistic Retrieval.
Results from research in information retrieval have suggested that significant improvements in retrieval effectiveness can be obtained by combining results from multiple index representations, query formulations, and search strategies. The inference net model of retrieval, which was designed from this point of view, treats information retrieval as an evidential reasoning process where multiple sources of evidence about document and query content are combined to estimate relevance probabilities. In this paper, we use a system based on this model to study the retrieval effectiveness benefits of combining the types of document and query information that are found in typical commercial databases and information services. The results indicate that substantial real benefits are possible.
Lab Report Special Section The University of Massachusetts Center for Intelligent Information Retrival.

Proceedings of the 17th Annual International ACM%2DSIGIR Conference on Research and Development in Information Retrieval. Dublin

Supporting Full%2DText Information Retrieval with a Persistent Object Store.
Full-text information retrieval systems have unusual and challenging data management requirements. Attempts have been made to satisfy these requirements using traditional (e.g., relational) database management systems. Those attempts, however, have produced rather discouraging results. Instead, information retrieval systems typically use custom data management facilities that require significant development effort and usually do not provide all of the services available from a standard database management system. Advanced data management systems, such as object-oriented database management systems and persistent object stores, offer a reasonable alternative to the two previous approaches. We have taken an existing information retrieval system (INQUERY) and substituted a persistent object store (Mneme) for the portion of the custom data management system that manages an inverted file index. The result is an improvement in performance and significant opportunities for the inform...
Document Retrieval and Routing Using the INQUERY System.
eval. We therefore used only two-word phrases. Sequences of more than two words were broken down into two-word subsequences. For example: (crude oil price trend) ) (crude oil), (oil price), (price trend). Using this modification of the original phrase extraction procedures, we were able to eliminate a number of special processing steps, and apply nounphrase -extraction processing to all sections.  We weighted the sections of the query as follows; title = 3n; description = 1n, narrative = 1; where n is the number of terms in the section. This was accomplished simply by duplicating the title three times, the description once, and by using the #SUM of the narrative terms.  1.2 PhraseFinder  PhraseFinder is a technique for corpus-based query expansion [8]. Phrasal concepts are identified using a part-of-speech tagger, and the text contexts of these concepts (words in close proximity) are used
Integrating Text and Database Systems Possibilities and Challenges.

Document Retrieval and Routing Using the INQUERY System.
eval. We therefore used only two-word phrases. Sequences of more than two words were broken down into two-word subsequences. For example: (crude oil price trend) ) (crude oil), (oil price), (price trend). Using this modification of the original phrase extraction procedures, we were able to eliminate a number of special processing steps, and apply nounphrase -extraction processing to all sections.  We weighted the sections of the query as follows; title = 3n; description = 1n, narrative = 1; where n is the number of terms in the section. This was accomplished simply by duplicating the title three times, the description once, and by using the #SUM of the narrative terms.  1.2 PhraseFinder  PhraseFinder is a technique for corpus-based query expansion [8]. Phrasal concepts are identified using a part-of-speech tagger, and the text contexts of these concepts (words in close proximity) are used

Fast Incremental Indexing for Full%2DText Information Retrieval.
Full-text information retrieval systems have traditionally been designed for archival environments. They often provide little or no support for adding new documents to an existing document collection, requiring instead that the entire collection be re-indexed. Modern applications, such as information filtering, operate in dynamic environments that require frequent additions to document collections. We provide this ability using a traditional inverted file index built on top of a persistent object store. The data management facilities of the persistent object store are used to produce efficient incremental update of the inverted lists. We describe our system and present experimental results showing superior incremental indexing and competitive query processing performance. Keywords: full-text document retrieval, incremental indexing, persistent object store, performance 1 Introduction Full-text information retrieval (IR) systems are well established tools for satisfying a user&#039;s inf...
Relevance Feedback and Inference Networks.
Relevance feedback, which modifies queries using judgements of the relevance of a few, highly-ranked documents, has historically been an important method for increasing the performance of information retrieval systems. In this paper, we extend the inference network model introduced by Turtle and Croft to include relevance feedback techniques. The difference between relevance feedback on text abstracts and full text collections is studied. Preliminary results for relevance feedback on the structured queries supported by the inference net model are also reported. 1 Introduction  Relevance feedback methods in information retrieval attempt to improve performance for a particular query by modifying the query, based on the user&#039;s reaction to the initial retrieved documents. Specifically, the user&#039;s judgements of the relevance or non-relevance of some of the documents retrieved are used to add new terms to the query and to reweight query terms. For example, if all the documents, that the user...
A Comparison of Indexing Techniques for Japanese Text Retrieval.

Effect of Multiple Query Representations on Information Retrieval System Performance.
Five independently generated Boolean query formulations for ten different TREC topics were produced by ten different expert online searchers. These different formulations were grouped, and the groups, and combinations of them, were used as searches against the TREC test collection, using the INQUERY probabilistic inference network retrieval engine. Results show that progressive combination of query formulations leads to progressively improving retrieval performance. Results were compared against the performance of INQUERY natural language based queries, and in combination with them. The issue of recall as a performance measure in large databases was raised, since overlap between the searches conducted in this study, and the TREC-1 searches, was smaller than expected.

TREC%2D2 Routing and Ad%2DHoc Retrieval Evaluation using the INQUERY System.
hasizes retrieval based on combination of evidence. Different text representations (such as words, phrases, paragraphs, or manually assigned keywords) and different versions of the query (such as natural language and Boolean) can be combined in a consistent probabilistic framework. This type of &#034;data fusion&#034; has been known to be effective in the information retrieval context for a number of years, and was one of the primary motivations for developing the inference net approach. Another feature of the inference net approach is the ability to capture complex structure in the network representing the information need (i.e. the query). A practical consequence  of this is that complex Boolean queries can be evaluated as easily as natural language queries and produce ranked output. It is also possible to represent &#034;rule-based&#034; or &#034;concept-based&#034; queries in the same probabilistic framework. This has led to us concentrating on automatic analysis of queries and techniques for enhancing
Guest Editor%27s Introduction.

Knowledge%2DBased and Statistical Approaches to Text Retrieval.

Retrieval Strategies for Hypertext.

The INQUERY Retrieval System.
As larger and more heterogeneous text databases become available, information retrieval research will depend on the development of powerful, efficient and flexible retrieval engines. In this paper, we describe a retrieval system (INQUERY) that is based on a probabilistic retrieval model and provides support for sophisticated indexing and complex query formulation. INQUERY has been used successfully with databases containing nearly 400,000 documents. 1 Introduction  The increasing interest in sophisticated information retrieval (IR) techniques has led to a number of large text databases becoming available for research. The size of these databases, both in terms of the number of documents in them, and the length of the documents that are typically full text, has presented significant challenges to IR researchers who are used to experimenting with two or three thousand document abstracts. In order to carry out research with different types of text representations, retrieval models, learni...
Retrieval of Complex Objects.
Traditional Information Retrieval (IR) considers documents as atomic units. In this paper, we show the retrieval of the components of the documents which satisfy best the information need. This finer granularity eases the browsing of the retrieval result. The approach supports multimedia and networked IR since multimedia documents are composed of other objects and networks combine several collections comprising the documents. We gain a unified view on networks, databases, and multimedia documents by considering them as complex objects --- retrieval among a heterogeneous document corpus can be modeled appropriately. We present a probabilistic retrieval function where the initial estimation of probabilistic parameters is based on the logical structure of documents and the retrieval process is described as probabilistic logical inference. Probabilistic parameters and the retrieval process are represented in probabilistic Datalog programs which are executed by HySpirit --- a system for pro...
A Loosely%2DCoupled Integration of a Text Retrieval System and an Object%2DOriented Database System.
Modern system development environments usually deploy the object management facilities of a so-called repository to store the documents created and maintained during system development. PCTE is the ISO and ECMA standard for a public tool interface for an open repository [23]. In this paper we present document retrieval extensions for an OQLoriented query language for PCTE. The extensions proposed cover (1) pattern matching, (2) term based document retrieval with automatically generated document description vectors, (3) the flexible definition of what is addressed as a &#034;document&#034; in agiven query, and (4) the integration of these facilities into a CASE tool. Whereas the integration of pattern matching facilities into query languages has been addressed by other authors before, the main contribution of our approach is the homogeneous integration of term based document retrieval and the flexible definition of documents.  1 Introduction  Repository-based applications are in wide-spread use i...
Experience with Large Document Collections %28Panel%29.

TIPSTER Panel %2D The University of Massachusetts TIPSTER Project.

Lexical Ambiguity and Information Retrieval.
Lexical ambiguity is a pervasive problem in natural language processing. However, little  quantitative information is available about the extent of the problem, or about the impact  that it has on information retrieval systems. We report on an analysis of lexical  ambiguity in information retrieval test collections, and on experiments to determine  the utility of word meanings for separating relevant from non-relevant documents. The  experiments show that there is considerable ambiguity even in a specialized database. Word senses
Information Filtering and Information Retrieval Two Sides of the Same Coin%3F
sides of the same coin?
A Comparison of Text Retrieval Models.

The University of Massachusetts TIPSTER Project.
Search engines are a critical tool for intelligence analysis. A number of innovations for search have been introduced since research with an emphasis on analyst needs began in the TIPSTER project. For example, the Inquery search engine introduced support for specification of complex queries in a probabilistic inference network framework. Recent research on language modeling has led to the development of Indri, a search engine that combines the best features of inference nets and language modeling in an architecture designed for large-scale applications. In this paper, we describe the Indri system and show how the query language is designed to support modern language technologies. We also present results demonstrating that Indri is both effective and efficient. 1.
Summary of Database Research Activities at The University of Massachusetts

Efficient probabilistic Inference for text retrieval.

The Use of Phrases and Structured Queries in Information Retrieval.

Editorial.
calls &amp; calendarEDITORIAL
Evaluation of an Inference Network%2DBased Retrieval Model.
The use of inference networks to support document retrieval is introduced. A network-based retrieval model is described and compared to conventional probabilistic and Boolean models. The performance of a retrieval system based on the inference network model is evaluated and compared to performance with conventional retrieval models,
Hypertext and Information Retrieval What are the Fundamental Concepts%3F %28Panel%29.

Inference Networks for Document Retrieval.
The use of inference networks to support document retrieval is introduced. A network-basead retrieval model is described and compared to conventional probabilis-tic and Boolean models. 1
Experiments with Query Acquisition and Use in Document Retrieval Systems.
Im some recent experimental document retrieval systems, emphasis has been placed on the acquisition of a detailed model of the information need through interaction with the user. It has been argued that these “enhanced ” queries, in combination with relevance feedback, will improve retrieval performance. In this paper, we describe a study with the aim of evaluating how easily enhanced queries can be acquired from users and how effectively this additional knowledge can be used in retrieval. The results indicate that significant effectiveness benefits can be obtained through the acquisition of domain concepts related to query concepts, together with their level of importance to the information need. 1
Term Clustering of Syntactic Phrases.
Term clustering and syntactic phrase formation are methods for transforming natural language text. Both have had only mixed success as strategies for improving the quality of text representations for document retrieval. Since the strengths of these methods are complementary, we have explored combining them to produce superior representations. In this paper we discuss our implementation of a syntactic phrase generator, as well as our preliminary experiments with producing phrase clusters. These experiments show small improvements in retrieval effectiveness resulting from the use of phrase clusters, but it is clear that corpora much larger than standard information retrieval test collections will be required to thoroughly evaluate the use of this technique.
Visual Interaction Between End Users and Goal Based Systems.

Towards Intelligent Information Retrieval An Overview of IR Research at U. Mass.

Interpreting nominal compounds for information retrieval.
In the first chapter, a novel application of text mining techniques in sociology is presented. In the second chapter, a technique of text mining based on the extraction of the information from the document titles is described. In the third chapter, the use of a hierarchical dictionary for text mining tasks is discussed.
Interactive retrieval of complex documents.
s Service. Journal of Chemical Information and Computer Sciences 15.226--233.  BIBLIOGRAPHY 123 MACKINLAY, JOCK, 1986. Automatic design of graphical presentations. Stanford University dissertation. Technical Report Stan-CS-86-1038. MANBER, UDI, &amp; SUN WU. 1994. GLIMPSE: a tool to search through entire file systems. In Proceedings of the Winter 1994 USENIX Conference, 23--31, San Francisco, CA. MANN, WILLIAM C., &amp; SANDRA A. THOMPSON. 1987. Rhetorical structure theory: A theory of text organization. Technical Report ISI/RS 87-190, ISI. MANNING, CHRISTOPHER D. 1993. Automatic acquisition of a large subcategorization dictionary from corpora. In Proceedings of the 31st Annual Meeting of the Association for Computational Lingusitics, 235--242, Columbus, OH. MARCHIONINI, GARY, PETER LIEBSCHER, &amp; XIA LIN. 1991. Authoring hyperdocuments: Designing for interaction. In Interfaces for information retrieval and online systems,  ed. by Martin Dillon, 119--131. New York, NY: Greenwood Press. MARKEY, ...
Chairman%27s Message.

A Retrieval Model for Incorporating Hypertext Links.
information. Instead one has to cope with the problem of different theories and typological approaches competing with each other of time. Analogies in architecture depend more on the perception devoted to some theoretical approach than on physical properties of an architectural case.  Architects use old solutions as source for inspiration in the design process. We present a model for the representation of architectural design cases. We face the problem of formalizing ill-structured domain knowledge without a consistent theoretical base. We further investigate the relationship between architectural design information and the languages used to represent that information. The theoretical model of language-game abstractions (LGAs) is presented as a means to study these relationships. Based on that model we propose an information structure that supports intelligent reasoning. This is used to retrieve appropriate case material about some design goal. The definition of similarity between desi...
Word Sense Disambiguation Using Machine%2DReadable Dictionaries.
In this paper, we describe a means for automatically building very large neural networks  (VLNNs) from definition texts in machine-readable dictionaries, and demonstrato the use of  these networks for word sense disambiguation. Our method brings together two earlier,  independent approaches to word sense disambiguation: the use of machine-readable  dictionaries and spreading and activation models. The automatic construction of VLNNs  enables real-size experiments with neural networks for natural languageprocessing, which in  turn provides insight into their behavior and design and can lead to possible improvements.
Retrieving documents by plausible inference An experimental study.

Support for Browsing in an Intelligent Text Retrieval System.
texts using relevancy signatures. In Proceedings of the
Relating Human Knowledge of Tasks to the Requirements of Plan Libraries.

Chairman%27s Message.

Retrieving Documents by Plausible Inference A Preliminary Study.

Implementing Ranking Strategies Using Text Signatures.

Chairman%27s Message.

Reasoning about Exceptions during Plan Execution Monitoring.

An Approach to Natural Language Processing for Document Retrieval.
Document retrieval systems have been restricted, by the nature of the task, to techniques that can be used with large numbers of documents and broad domains. The most effective techniques that have been developed are based on the statistics of word occurrences in text. In this paper, we describe an approach to using natural language processing (NLP) techniques for what is essentially a natural language problem - the comparison of a request text with the text of document titles and abstracts. The proposed NLP techniques are used to develop a request model based on &#034;conceptual case frames&#034; and to compare this model with the texts of candidate documents. The request model is also used to provide information to statistical search techniques that identify the candidate documents. As part of a preliminary evaluation of this approach, case frame representations of a set of requests from the CACM collection were constructed. Statistical searches carried out using dependency and relative import...
Supporting Offics Document Architectures with Constrained Types.

Approaches to Intelligent Information Retrieval.
We have built a hybrid Case-Based Reasoning (CBR) and Information Retrieval (IR) system that generates a query to the IR system by using information derived from CBR analysis of a problem situation. The query is automatically formed by submitting in text form a set of highly relevant cases, based on a CBR analysis, to a modified version of INQUERY&#039;s relevance feedback module. This approach extends the reach of CBR, for retrieval purposes, to much larger corpora and injects knowledge-based techniques into traditional IR. 1 Introduction  One strength of Case-Based Reasoning (CBR) systems is the ability to reason about a problem case and perform highly intelligent problem-solving, such as the generation of legal arguments or detailed operational plans [9]. In particular, CBR systems have at their core the ability to retrieve highly relevant cases. However, CBR systems are limited by the availability of cases actually represented in their case bases. Among current case-based reasoning syst...
Approaches to intelligent information retrieval.
We have built a hybrid Case-Based Reasoning (CBR) and Information Retrieval (IR) system that generates a query to the IR system by using information derived from CBR analysis of a problem situation. The query is automatically formed by submitting in text form a set of highly relevant cases, based on a CBR analysis, to a modified version of INQUERY&#039;s relevance feedback module. This approach extends the reach of CBR, for retrieval purposes, to much larger corpora and injects knowledge-based techniques into traditional IR. 1 Introduction  One strength of Case-Based Reasoning (CBR) systems is the ability to reason about a problem case and perform highly intelligent problem-solving, such as the generation of legal arguments or detailed operational plans [9]. In particular, CBR systems have at their core the ability to retrieve highly relevant cases. However, CBR systems are limited by the availability of cases actually represented in their case bases. Among current case-based reasoning syst...
Distributed Expert%2DBased Information Systems An Interdisciplinary Approach.

User%2DSpecified Domain Knowledge for Document Retrieval.

An expert assistant for document retrieval system.

A Knowledge%2DBased Approach to Data Management for Intelligent User Interfaces.

Task Management for an Intelligent Interface.
DECEMBER 1985 VOL. 8 NO. 4 a quarterly bulletin of the IEEE computer society
A comparison of a network structure and a database system used for document retrieval.

The Use of Adaptive Mechanisms for Selection of Search Strategies in Document Retrieval Systems.

Task Support in an Office System.

The Role of Context and Adaptation in User Interfaces.

A Network Organization Used for Document Retrieval.
A network organization for implementing a document retrieval system is proposed. This organization has significant advantages in terms of the range of searches that can be used when compared to either inverted or clustered file organizations. Algorithms for generating and maintaining the network are described together with experiments designed to test their efficiency and effectiveness. I. Introduction physically grouping documents that are likely to The main types of file organlzd~1on tha ~ b ~ retrieved by the same queries. nave been proposed for the implementation of In this paper, we investigate a network document retrleval systems rare inverted files organizatlon which combines Lll ~ advantages of and c±ustered files [SALT82J. Although other inverted and clustered files. This organization
Applications for Information Retrieval Techniques in the Office.

The Implementation of a Document Retrieval System.
A network organization for implementing a document retrieval system is proposed. This organization has significant advantages in terms of the range of searches that can be used when compared to either inverted or clustered file organizations. Algorithms for generating and maintaining the network are described together with experiments designed to test their efficiency and effectiveness. I. Introduction physically grouping documents that are likely to The main types of file organlzd~1on tha ~ b ~ retrieved by the same queries. nave been proposed for the implementation of In this paper, we investigate a network document retrleval systems rare inverted files organizatlon which combines Lll ~ advantages of and c±ustered files [SALT82J. Although other inverted and clustered files. This organization
Experiments with Automatic Text Filing and Retrieval in the Office Environment.

Incorporating Different Search Models into One Document Retrieval System.

A model of cluster searching bases on classification.

On the Implementation of Some Models of Document Retrieval.

A File Organization for Cluster%2DBased Retrieval.

An evaluation of Goffman%27s indirect retrieval method.

Document clustering An evaluation of some experiments with the cranfield 1400 collection.

