Method of Semantic Refinement for Enterprise Search

Alexey Pismak, Serge Klimenkov, Eugeny Tsopa, Alexandr Yarkeev, Vladimir Nikolaev, Anton Gavrilov

Abstract

In this paper, we propose an approach of using the semantic refinement of the input search query for the enterprise search systems. The problem of enterprise search is actual because of the amount of processed data. Even with a good organization of documents, the process of searching for specific documents or specific data in these documents is very laborious. But even more significant problem is that the required content may have the matching meaning, but expressed with different words in the different languages, which prevents it from appearing in the search result. The proposed approach uses semantic refinement of the search query. First, the concepts are extracted from the semantic network based on translingual lexemes of the user query string, allowing to perform the search based on the senses rather than word forms. In addition, several rules are applied to the query in order to include or exclude senses which can affect the relevance and the pertinence of the search result.

Download


Paper Citation