Laplacian Co%2Dhashing of Terms and Documents.

Self%2Dtaught hashing for fast similarity search.

Self%2DTaught Hashing for Fast Similarity Search

Proceeding of the ACM First International Workshop on Complex Networks Meet Information %26 Knowledge Management

Bridging the gap complex networks meet information and knowledge management.

Time%2DSensitive Language Modelling for Online Term Recurrence Prediction.
Abstract. We address the problem of online term recurrence prediction: for a stream of terms, at each time point predict what term is going to recur next in the stream given the term occurrence history so far. It has many applications, for example, in Web search and social tagging. In this paper, we propose a time-sensitive language modelling approach to this problem that effectively combines term frequency and term recency information, and describe how this approach can be implemented efficiently by an online learning algorithm. Our experiments on a real-world Web query log dataset show significant improvements over standard language modelling. 1
Batch%2DMode Computational Advertising Based on Modern Portfolio Theory.
Abstract. The research on computational advertising so far has focused on finding the single best ad. However, in many real situations, more than one ad can be presented. Although it is possible to address this problem myopically by using a single-ad optimisation technique in serial-mode, i.e., one at a time, this approach can be ineffective and inefficient because it ignores the correlation between ads. In this paper, we make a leap forward to address the problem of finding the best ads in batch-mode, i.e., assembling the optimal set of ads to be presented altogether. The key idea is to achieve maximum revenue while controlling the level of risk by diversifying the set of ads. We show how the Modern Portfolio Theory can be applied to this problem to provide elegant solutions and deep insights. 1
What queries are likely to recur in web search%3F

The recurrence dynamics of social tagging.
Social tagging systems have become increasingly popular for sharing and organizing web resources. Tag recommendation is a common feature of social tagging systems. Social tagging by nature is an incremental process, meaning that once a user has saved a web page with tags, the tagging system can provide more accurate predictions for the user, based on the user’s incremental behavior. However, existing tag prediction methods do not consider this important factor, in which their training and test datasets are either split by a fixed time stamp or randomly sampled from a larger corpus. In our temporal experiments, we perform a time-sensitive sampling on an existing public dataset, resulting in a new scenario which is much closer to “real-world”. In this paper, we address the problem of tag prediction by proposing a probabilistic model for personalized tag prediction. The model is a Bayesian approach, and integrates three factors— an ego-centric effect, environmental effects and web page content. Two methods—both intuitive calculation and learning optimization—are provided for parameter estimation. Pure graphbased methods which may have significant constraints (such as every user, every item and every tag has to occur in at least p posts) cannot make a prediction in most “real world ” cases while our model improves the F-measure by over 30 % compared to a leading algorithm on a publicly-available real-world dataset.
Classifying networked entities with modularity kernels.
Statistical machine learning techniques for data classification usually assume that all entities are i.i.d. (independent and identically distributed). However, real-world entities often interconnect with each other through explicit or implicit relationships to form a complex network. Although some graph-based classification methods have emerged in recent years, they are not really suitable for complex networks as they do not take the degree distribution of network into consideration. In this paper, we propose a new technique, Modularity Kernel, that can effectively exploit the latent community structure of networked entities for their classification. A number of experiments on hypertext datasets show that our proposed approach leads to excellent classification performance in comparison with the state-of-the-art methods.
Extracting community structure features for hypertext classification.
Standard text classification techniques assume that all documents are independent and identically distributed (i.i.d.). However, hypertext documents such as web pages are interconnected with links. How to take advantage of such links as extra evidences to enhance automatic classification of hypertext documents is a non-trivial problem. We think a collection of interconnected hypertext documents can be considered as a complex network, and the underlying community structure of such a document network contains valuable clues about the right classification of documents. This paper introduces a new technique, Modularity Eigenmap, that can effectively extract community structure features from the document network which is induced from document link information only or constructed by combining both document content and document link information. A number of experiments on real-world benchmark datasets show that the proposed approach leads to excellent classification performance in comparison with the state-of-the-art methods. 1
A fast approximate algorithm for large%2Dscale Latent Semantic Indexing.
Latent Semantic Indexing (LSI) is an effective method to discover the underlying semantic structure of data. It has numerous applications in information retrieval and data mining. However, the computational complexity of LSI may be prohibitively high when applied to very large datasets. In this paper, we present a fast approximate algorithm for large-scale LSI that is conceptually simple and theoretically justified. Our main contribution is to show that the proposed algorithm has provable error bound and linear computational complexity. 1
Learning classifiers without negative examples A reduction approach.
The problem of PU Learning, i.e., learning classifiers with positive and unlabelled examples (but not negative examples), is very important in information retrieval and data mining. We address this problem through a novel approach: reducing it to the problem of learning classifiers for some meaningful multivariate performance measures. In particular, we show how a powerful machine learning algorithm, Support Vector Machine, can be adapted to solve this problem. The effectiveness and efficiency of the proposed approach have been confirmed by our experiments on three real-world datasets. 1
Learning with support vector machines for query%2Dby%2Dmultiple%2Dexamples.

Learning to Separate Text Content and Style for Classification.

Extracting key%2Dsubstring%2Dgroup features for text classification.
In many text classification applications, it is appealing to take every document as a string of characters rather than a bag of words. Previous research studies in this area mostly focused on different variants of generative Markov chain models. Although discriminative machine learning methods like Support Vector Machine (SVM) have been quite successful in text classification with word features, it is neither effective nor efficient to apply them straightforwardly taking all substrings in the corpus as features. In this paper, we propose to partition all substrings into statistical equivalence groups, and then pick those groups which are important (in the statistical sense) as features (named keysubstring-group features) for text classification. In particular, we propose a suffix tree based algorithm that can extract such features in linear time (with respect to the total number of characters in the corpus). Our experiments on English, Chinese and Greek datasets show that SVM with key-substring-group features can achieve outstanding performance for various text classification tasks.
Text classification with kernels on the multinomial manifold.

Web Based Question Answering with Aggregation Strategy.
Abstract. The Web is apparently an ideal source of answers to a large variety of questions, due to the tremendous amount of information available online. This paper describes a question answering system LAMP, which can instantly find answers from the Web. LAMP is publicly accessible on the Web, and its performance is comparable to the state-of-the-art question answering systems. One novel characteristic of this system is that its answer selection method is based on aggregation, which is shown to be more effective than the common methods based on individual or redundancy. 1
Semantic%2C Hierarchical%2C Online Clustering of Web Search Results.

Web Taxonomy Integration Using Spectral Graph Transducer.

Web Taxonomy Integration Using Spectral Graph Transducer.

Web taxonomy integration through co%2Dbootstrapping.

Web taxonomy integration through co%2Dbootstrapping.



Web taxonomy integration using support vector machines.
We address the problem of integrating objects from a source taxonomy into a master taxonomy. This problem is not only currently pervasive on the web, but also important to the emerging semantic web. A straightforward approach to automating this process would be to train a classifier for each category in the master taxonomy, and then classify objects from the source taxonomy into these categories. In this paper we attempt to use a powerful classification method, Support Vector Machine (SVM), to attack this problem. Our key insight is that the availability of the source taxonomy data could be helpful to build better classifiers in this scenario, therefore it would be beneficial to do transductive learning rather than inductive learning, i.e., learning to optimize classification performance on a particular set of test examples. Noticing that the categorizations of the master and source taxonomies often have some semantic overlap, we propose a method, Cluster Shrinkage (CS), to further enhance the classification by exploiting such implicit knowledge. Our experiments with real-world web data show substantial improvements in the performance of taxonomy integration.
Learning to integrate web taxonomies.
We investigate machine learning methods for automatically integrating objects from different taxonomies into a master taxonomy. This problem is not only currently pervasive on the Web, but is also important to the emerging Semantic Web. A straightforward approach to automating this process would be to build classifiers through machine learning and then use these classifiers to classify objects from the source taxonomies into categories of the master taxonomy. However, conventional machine learning algorithms totally ignore the availability of the source taxonomies. In fact, source and master taxonomies often have common categories under different names or other more complex semantic overlaps. We introduce two techniques that exploit the semantic overlap between the source and master taxonomies to build better classifiers for the master taxonomy. The first technique, Cluster Shrinkage, biases the learning algorithm against splitting source categories by making objects in the same category appear more similar to each other. The second technique, Co-Bootstrapping, tries to facilitate the exploitation of inter-taxonomy relationships by providing category indicator functions as additional features for the objects. Our experiments with real-world Web data show that these proposed add-on techniques can enhance various machine learning algorithms to achieve substantial improvements in performance for taxonomy integration.
Question classification using support vector machines.
Question classification is very important for question answering. This paper presents our research work on automatic question classification through machine learning approaches. We have experimented with five machine learning algorithms: Nearest
A Language Modeling Approach to Passage Question Answering.
Passage retrieval is an essential part of question answering systems. In this paper we use statistical language models to perform this task. Previous work has shown that language modeling techniques provide better results for both, document and passage retrieval. The motivation behind this paper is to define new smoothing methods for passage retrieval in question answering systems. The long term objective is to improve the quality of question answering systems to isolate the correct answer by choosing and evaluating the appropriate section of a document. In this work we use a three step approach. The first two steps are standard document and passage retrieval using the Lemur toolkit. As a novel contribution we propose as the third step a re-ranking using dedicated backing-off distributions. In particular backing-off from the passage-based language model to a language model trained on the document from which the passage is taken shows a significant improvement. For a TREC question answering task we can increase the mean average precision from 0.127 to
Web Based Pattern Mining and Matching Approach to Question Answering.
We describe herein a Web based pattern mining and matching approach to question answering. For each type of questions, a lot of textual patterns can be learned from the Web automatically, using the TREC QA track data as training examples. These textual patterns are assessed by the concepts of support and confidence, which are borrowed from the data mining community. Given a new unseen question, these textual patterns can be utilized to extract and rank the plausible answers on the Web. The performance of this approach has been evaluated also by the TREC QA track data.
A novel Web usage mining approach for search engines.

An efficient algorithm to rank Web resources.
How to rank Web resources is critical to Web Resource Discovery (Search Engine). This paper not only points out the weakness of current approaches, but also presents in-depth analysis of the multidimensionality and subjectivity of rank algorithms. From a dynamics viewpoint, this paper abstracts a user&#039;s Web surfing action as a Markov model. Based on this model, we propose a new rank algorithm. The result of our rank algorithm, which synthesizes the relevance, authority, integrativity and novelty of each Web resource, can be computed efficiently not by iteration but through solving a group of linear equations.  2000 Published by Elsevier Science B.V. All rights reserved.
A Data Model and Algebra for the Web.

An Object Oriented Data Model for Web and Its Algebra.

