Mining and modeling linkage information from citation context for improving biomedical literature retrieval.

Brain Informatics

Proceedings of the 19th ACM Conference on Information and Knowledge Management

Brain Informatics

Revisiting Rocchio%27s Relevance Feedback Algorithm for Probabilistic Models.

Exploring Social Annotation Tags to Enhance Information Retrieval Performance.

Enhancing Content%2DBased Image Retrieval Using Machine Learning Techniques.

Optimal IR How Far Away%3F

Promoting Ranking Diversity for Biomedical Information Retrieval Using Wikipedia.

Medical search and classification tools for recommendation.

S%2DPLASA  adaptive sentiment analysis with application to sales performance prediction.

A survival modeling approach to biomedical search result diversification using wikipedia.

A quality%2Daware model for sales prediction using reviews.

Passage extraction and result combination for genomics information retrieval.

Towards a Better Ranking for Biomedical Information Retrieval Using Context.

Integrating multiple document features in language models for expert finding.

Towards a Better Ranking for Biomedical Information Retrieval Using Context.

BioCLink A Probabilistic Approach for Improving Genomics Search with Citation Links.

Discovering Temporal Associations among Significant Changes in Gene Expression.

Diverging patterns discovering significant frequency change dissimilarities in large databases.



Boosting Biomedical Information Retrieval Performance through Citation Graph An Empirical Study.

A bayesian learning approach to promoting diversity in ranking for biomedical information retrieval.

A graph%2Dbased approach to mining multilingual word associations from wikipedia.

TREC%2DCHEM large scale chemical information retrieval evaluation at TREC.

Modeling document features for expert finding.

Integration of Genomic

Machine Learning Approaches to Information Retrieval and its Applications to the Web

Modeling and Predicting the Helpfulness of Online Reviews.

A Dynamic Window Based Passage Extraction Algorithm for Genomics Information Retrieval.

A reranking model for genomics aspect search.

York University at TREC 2008 Blog Track.

HelpMeter A Nonlinear Model for Predicting the Helpfulness of Online Reviews.



York University at TREC 2007 Enterprise Document Search.

York University at TREC 2007 Genomics Track.

Knowledge Retrieval %28KR%29.
Where is the Life we have lost in living? Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information? — T.S. Eliot, The Rock, 1934. With the ever-increasing growth of data and information, finding the right knowledge becomes a real challenge and an urgent task. Traditional data and information retrieval systems that support the current web are no longer adequate for knowledge seeking tasks. Knowledge retrieval systems will be the next generation of retrieval system serving those purposes. Basic issues of knowledge retrieval systems are examined and a conceptual framework of such systems is proposed. Theories and Technologies such as theory of knowledge, machine learning and knowledge discovery, psychology, logic and inference, linguistics, etc. are briefly mentioned for the implementation of knowledge retrieval systems. Two applications of knowledge retrieval in rough sets and biomedical domains are presented. 1.
Comparison of Interestingness Measures for Web Usage Mining an Empirical Study.

Finding molecular complexes through multiple layer clustering of protein interaction networks.

Comparison of Interestingness Measures for Web Usage Mining an Empirical Study.

An effective extension to okapi for biomedical text mining.

A multilevel searching and re%2Dranking framework for information retrieval.

Applying Data Mining to Pseudo%2DRelevance Feedback for High Performance Text Retrieval.
In this paper, we investigate the use of data mining, in particular the text classification and co-training techniques, to identify more relevant passages based on a small set of labeled passages obtained from the blind feedback of a retrieval system. The data mining results are used to expand query terms and to re-estimate some of the parameters used in a probabilistic weighting function. We evaluate the data mining based feedback method on the TREC HARD data set. The results show that data mining can be successfully applied to improve the text retrieval performance. We report our experimental findings in detail. 1
Mining and Modeling Database User Access Patterns.

Boosting Prediction Accuracy on Imbalanced Datasets with SVM Ensembles.
Abstract. Learning from imbalanced datasets is inherently difficult due to lack of information about the minority class. In this paper, we study the performance of SVMs, which have gained great success in many real applications, in the imbalanced data context. Through empirical analysis, we show that SVMs suffer from biased decision boundaries, and that their prediction performance drops dramatically when the data is highly skewed. We propose to combine an integrated sampling technique with an ensemble of SVMs to improve the prediction performance. The integrated sampling technique combines both over-sampling and under-sampling techniques. Through empirical study, we show that our method outperforms individual SVMs as well as several other state-of-the-art classifiers. 1
Boosting Prediction Accuracy on Imbalanced Datasets with SVM Ensembles.
Abstract. Learning from imbalanced datasets is inherently difficult due to lack of information about the minority class. In this paper, we study the performance of SVMs, which have gained great success in many real applications, in the imbalanced data context. Through empirical analysis, we show that SVMs suffer from biased decision boundaries, and that their prediction performance drops dramatically when the data is highly skewed. We propose to combine an integrated sampling technique with an ensemble of SVMs to improve the prediction performance. The integrated sampling technique combines both over-sampling and under-sampling techniques. Through empirical study, we show that our method outperforms individual SVMs as well as several other state-of-the-art classifiers. 1
Concept%2Dbased biomedical text retrieval.

A platform for Okapi%2Dbased contextual information retrieval.

York University at TREC 2006 Enterprise Email Discussion Search.

York University at TREC 2006 Genomics Track.

York University at TREC 2006 Legal Track.
York University participated in the legal track this year. For this track, we developed an Okapi-based Legal Search Engine (LSE) v1.0. Our experiments mainly focused on evaluating the effect of a probabilistic text retrieval model on the legal domain. In order to address the special problems in legal text retrieval, new automatic feedback methods and term weighting methods are proposed and tested. 1
Hierarchical grouping of association rules and its application to a real%2Dworld domain.

Applying language modeling to session identification from database trace logs.

Finding and Analyzing Database User Sessions.

A Machine Learning Approach to Identifying Database Sessions Using Unlabeled Data.


Using contextual information to improve retrieval performance.


A Distance%2DBased Algorithm for Clustering Database User Sessions.

A dual index model for contextual information retrieval.

York University at TREC 2005 Genomics Track.
Our Genomics experiments mainly focus on addressing three major problems in biomedical  information retrieval. The three problems are: (1) how to deal with synonyms? (2) how to deal  with the frequent use of acronyms? (3) how to deal with homonyms? In particular, we propose  two query expansion algorithms to construct structured queries for our experiments. The mean  average precision (MAP) for our automatic run &#034;york05ga1&#034; using Algorithm 1 was 0.2888 and  for our manual run &#034;york05gm1&#034; using Algorithm 2 was 0.3020. The evaluation results show  that both algorithms are e#ective for improving retrieval performance. We also find that some  other techniques such as pseudo-relevance feedback and using an extended stop word set can  make positive contributions to the retrieval performance.
York University at TREC 2005 SPAM Track.
We propose a variant of the k-nearest neighbor classification method, called  instance-weighted k-nearest neighbor method, for adaptive spam filtering. The method  assigns two weights, distance weight and correctness weight, to a training instance, and  makes use of the two weights when classifying a new email. The correctness weight is also  used in the maintenance of the training data to make the training data more adaptive to the  changes of spam characteristics. We submitted 4 spam filters to the Spam Track. Two of  the filters are purely based on the instance-weighted kNN method. The two other filters  combine the kNN method with other spam filtering and classification techniques. We report  the official results of our submissions on the Spam Track evaluation data sets.
York University at TREC 2005 HARD Track.

York University at TREC 2005 SPAM Track.
We propose a variant of the k-nearest neighbor classification method, called  instance-weighted k-nearest neighbor method, for adaptive spam filtering. The method  assigns two weights, distance weight and correctness weight, to a training instance, and  makes use of the two weights when classifying a new email. The correctness weight is also  used in the maintenance of the training data to make the training data more adaptive to the  changes of spam characteristics. We submitted 4 spam filters to the Spam Track. Two of  the filters are purely based on the instance-weighted kNN method. The two other filters  combine the kNN method with other spam filtering and classification techniques. We report  the official results of our submissions on the Spam Track evaluation data sets.
York University at TREC 2005 Terabyte Track.

York University at TREC 2004 HARD and Genomics Tracks.

Clustering Web Surfers with Probabilistic Models in a Real Application.

Dynamic Web log session identification with statistical language models.
We present a novel session identification method based on statistical language modeling.
Feature Selection with Rough Sets for Web Page Classification.
Abstract. Web page classification is the problem of assigning predefined categories to web pages. A challenge in web page classification is how to deal with the high dimensionality of the feature space. We present a feature reduction method based on the rough set theory and investigate the effectiveness of the rough set feature selection method on web page classification. Our experiments indicate that rough set feature selection can improve the predictive performance when the original feature set for representing web pages is large. 1


Objective and Subjective Algorithms for Grouping Association Rules.
We propose two algorithms for grouping and summarizing association rules. The first algorithm recursively groups rules according to the structure of the rules and generates a tree of clusters as a result. The second algorithm groups the rules according to the semantic distance between the rules by making use of an autometically tagged semantic tree-structured network of items. We provide a case study in which the proposed algorithms are evaluated. The results show that our grouping methods are effective and produce good grouping results. 1
Text classification in Asian languages without word segmentation.
We present a simple approach for Asian  language text classification without word  segmentation, based on statistical n-gram  language modeling. In particular, we examine  Chinese and Japanese text classification.
Applying Machine Learning to Text Segmentation for Information Retrieval.
We propose a self-supervised word segmentation technique for text segmentation in Chinese  information retrieval. This method combines the advantages of traditional dictionary  based, character based and mutual information based approaches, while overcoming many of  their shortcomings. Experiments on TREC data show this method is promising. Our method  is completely language independent and unsupervised, which provides a promising avenue  for constructing accurate multi-lingual or cross-lingual information retrieval systems that are  exible and adaptive. We  nd that although the segmentation accuracy of self-supervised  segmentation is not as high as some other segmentation methods, it is enough to give comparable  (in some cases even better) retrieval performance. It is commonly believed that word  segmentation accuracy is monotonically related to retrieval performance in Chinese information  retrieval. However, for Chinese, we  nd that the relationship between segmentation and  retrieval performance is in fact nonmonotonic; that is, at around 70% word segmentation  accuracy an over-segmentation phenomenon begins to occur which leads to a reduction in  information retrieval performance. We demonstrate this eect by presenting an empirical  investigation of information retrieval on Chinese TREC data, using a wide variety of word  segmentation algorithms with word segmentation accuracies ranging from 44% to 95%, including  70% word segmentation accuracy from our self-supervised word-segmentation approach.
Organizing knowledge and political process reconfiguration in the context of BPR.

Comparison of interestingness functions for learning web usage patterns.

Investigating the Relationship between Word Segmentation Performance and Retrieval Performance in Chinese IR.
It is commonly believed that word segmentation accuracy is monotonically related to retrieval performance in Chinese information retrieval. In this paper we show that, for Chinese, the relationship between segmentation and retrieval performance is in fact nonmonotonic; that is, at around 70% word segmentation accuracy an over-segmentation phenomenon begins to occur which leads to a reduction in information retrieval performance. We demonstrate this e ect by presenting an empirical investigation of information retrieval on Chinese TREC data, using a wide variety of word segmentation algorithms with word segmentation accuracies ranging from 44% to 95%. It appears that the main reason for the drop in retrieval performance is that correct compounds and collocations are preserved by accurate segmenters, while they are broken up by less accurate (but reasonable) segmenters, to a surprising advantage. This suggests that words themselves might be too broad a notion to conveniently capture the general semantic meaning of Chinese text.
Discovery of Interesting Association Rules from Livelink Web Log Data.

Using self%2Dsupervised word segmentation in Chinese information retrieval.
We propose a self-supervised word-segmentation technique for Chinese information retrieval. This method combines the advantages of traditional dictionary based approaches with character based approaches, while overcoming many of their shortcomings. Experiments on TREC data show comparable performance to both the dictionary based and the character based approaches. However, our method is language independent and unsupervised, which provides a promising avenue for constructing accurate multilingual information retrieval systems that are flexible and adaptive.
Probability%2DBased Chinese Text Processing and Retrieval.

A Case Study for Learning from Imbalanced Data Sets.

Probability%2DBased Chinese Text Processing and Retrieval.

Okapi Chinese Text Retrieval Experiments at TREC%2D6.
Introduction  The focus of the Okapi TREC#6 Chinese experiments is on investigating the e#ectiveness of di#erent automatic indexing methods and phrase weighting for retrieval based on probabilistic models over Chinese text. We compare di#erent probabilistic weighting methods based on a range of word and single character approaches.  There are two indexing methods used in our experiments. One indexing method is to use linguistic units #words, compound words and phrases# in texts as indexing terms to represent the texts. We refer to this method as the word approach. For this approach, text segmentation, which divides text into linguistic units, is regarded not only as a necessary precursor but also as a bottleneck of this kind of system #1#. The other method for indexing texts is based on single Chinese characters, in which texts are indexed bythecharacters appearing in the texts #2#. By using single character approaches, a search could be conducted for anymulti-character word or 
Okapi at TREC%2D5.
Introduction: summary  General  City submitted two runs each for the automatic ad hoc, very large collection track, automatic routing and Chinese track; and took part in the interactive and #ltering tracks. There were no very signi#cant new developments; the same Okapi-style weighting as in TREC#3 and TREC#4 was used this time round, although there were attempts, in the ad hoc and more notably in the Chinese experiments, to extend the weighting to cover searches containing both words and phrases. All submitted runs except for the Chinese incorporated run-time passage determination and searching. The Okapi back-end search engine has been considerably speeded, and a few new functions incorporated. See Section 3.  Automatic ad hoc #and VLC track#  After a good deal of rather fruitless exploration, including further work with two-term phrases, the method #nally used was much the same as in TRECs 3 and 4: expansion using terms from the top documents retrieved by 
