Abstract | ||
---|---|---|
In this paper, a novel sentence retrieval model with type-based expansion is proposed. In this retrieval model, sentences expected to be relevant should meet with the requirements both in query terms and query types. To obtain the information about query types, this paper proposes a solution based on classification, which utilizes the potential associations between terms and information types to obtain the optimized classification results. Inspired by the idea that relevant sentences always tend to occur nearby, this paper further re-ranks each sentence by considering the relevance of its adjacent sentences. The proposed retrieval model has been compared with other traditional retrieval models and experiment results indicate its significant improvements in retrieval effectiveness. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1007/978-3-540-72584-8_91 | International Conference on Computational Science (1) |
Keywords | Field | DocType |
information type,query type,sentence retrieval,adjacent sentence,novel sentence retrieval model,retrieval effectiveness,proposed retrieval model,query term,type-based query expansion,optimized classification result,retrieval model,traditional retrieval model,query expansion | Divergence-from-randomness model,Query language,Query expansion,Information retrieval,Computer science,Web query classification,Ranking (information retrieval),Natural language processing,Artificial intelligence,Term Discrimination,Concept search,Visual Word | Conference |
Volume | ISSN | Citations |
4487 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 11 | 4 |