Browsing by Author "Liang, T"

Now showing 1 - 5 of 5

A linguistic approach to a Chinese query model
(2001-01-01)
Liang, T
;
Wu, TC
;
資訊工程學系
;
Department of Computer Science
As network and input technologies have improved rapidly, researches in information retrieval have become more important than ever. An user-friendly and efficient query interface will certainly facilitate utilization of an information retrieval system. In this paper, a linguistic approach to Chinese query models is proposed and investigated. It includes Chinese linguistic variables, quantifiers and combination queries. Meanwhile, an improved computation method is proposed to handle general conjunctive queries, and it is proved to be better than traditional fuzzy computation in terms of a higher level of retrieval satisfaction. Experimental results also verify that the proposed quantifiers do not only simplify a multi-criteria request. bur also yield a higher level of retrieval satisfaction than do Boolean queries.
Anaphora resolution for biomedical literature by exploiting multiple resources
(2005)
Liang, T
;
Lin, YH
;
資訊工程學系
;
Department of Computer Science
In this paper, a resolution system is presented to tackle nominal and pronominal anaphora in biomedical literature by using rich set of syntactic and semantic features. Unlike previous researches, the verification of semantic association between anaphors and their antecedents is facilitated by exploiting more outer resources, including UMLS, WordNet, GENIA Corpus 3.02p and PubMed. Moreover, the resolution is implemented with a genetic algorithm on its feature selection. Experimental results on different biomedical corpora showed that such approach could achieve promising results on resolving the two common types of anaphora.
Empirical textual mining to protein entities recognition from PubMed corpus
(2005)
Liang, T
;
Shih, PK
;
資訊工程學系
;
Department of Computer Science
Named Entity Recognition (NER) from biomedical literature is crucial in biomedical knowledge base automation. In this paper, both empirical rule and statistical approaches to protein entity recognition are presented and investigated on a general corpus GENIA 3.02p and a new domain-specific corpus SRC. Experimental results show the rules derived from SRC are useful though they are simpler and more general than the one used by other rule-based approaches. Meanwhile, a concise HMM-based model with rich set of features is presented and proved to be robust and competitive while comparing it to other successful hybrid models. Besides, the resolution of coordination variants common in entities recognition is addressed. By applying heuristic rules and clustering strategy, the presented resolver is proved to be feasible.
Optimal weight assignment for a Chinese signature file
(1996-03-01)
Liang, T
;
Lee, SY
;
Yang, WP
;
資訊科學與工程研究所
;
Institute of Computer Science and Engineering
The performance of a character-based Chinese text retrieval scheme (the combined scheme) is investigated. In the scheme both the monogram keys (singleton characters) and bigram keys (consecutive character pairs) are encoded into document signatures such that half of the bits in every signature are set. For disyllabic queries, an analytical expression of the false hit rate that accounts for both random false hits and adjacency false hits is proposed. Then optimal monogram and bigram weight assignments together with the corresponding minimal false hit rate are derived in terms of signature length, storage overhead of the combined scheme, and the occurrence frequency and the association value of a disyllabic query. The theoretical predictions of the optimal weight assignments and the minimal false hit rate are tested and verified in experiments using a real Chinese corpus for disyllabic queries of different association values. Satisfactory agreement between the experimental results and theoretical predictions is found.
Web-based unsupervised learning for query formulation in question answering
(2005)
Wang, YC
;
Wu, JC
;
Liang, T
;
Chang, JS
;
資訊工程學系
Converting questions to effective queries is crucial to open-domain question answering systems. In this paper, we present a web-based unsupervised learning approach for transforming a given natural-language question to an effective query. The method involves querying a search engine for Web passages that contain the answer to the question, extracting patterns that characterize fine-grained classification for answers, and linking these patterns with n-grams in answer passages. Independent evaluation on a set of questions shows that the proposed approach outperforms a naive keyword-based approach in terms of mean reciprocal rank and human effort.