Context-dependent opinion retrieval for high precision results at top documents

Document Type

Conference Proceeding

Publication Title

Proceedings of the 4th International Conference on Internet Technologies and Applications, ITA 11


Existing opinion retrieval techniques do not provide context-dependent relevant results. They are based on frequency of query terms, such that all documents containing query terms are retrieved, regardless of contextual relevance to the intent of the human seeking the opinion. This can be described as non-contextual relevance problem common to opinion retrieval systems such as Google Blogs Search and Technorati Blog Directory. We believe this problem is caused by the presence of large proportion of irrelevant textual contents within the retrieved documents. As a result of this problem, opinion retrieved for the purpose of qualitative analysis of products, performance measurement for companies, and public reactions to political decisions can be largely biased. Thus, we propose a sentence-level contextual model for opinion retrieval task using grammatical tree derivations and approval voting mechanism. Experiments conducted over a large scale Blog corpus show that the proposed approach gives context-dependent results with high precision at top documents. We believe the proposed technique can be used for expert search systems, faceted-opinion retrieval, opinion trend analytic, and personalized web search.

First Page


Last Page


Publication Date


This document is currently not available here.