Semantic Retrieval Based on User Intention Recognition in Engineering Domain

Abstract

In the big data era, the rapid explosion and diversification of knowledge have brought various difficulties to effective knowledge retrieval. The keyword matching-based retrieval can only carry out character matching mechanically, which ignores the semantic content contained in the keyword itself; ontology-based semantic retrieval does not recognize and analyze the structural relationship among keywords user inputs, so it is unable to mine the implicit user intentions; the effectiveness of the intention recognition method based on user’s behaviors is not easy to guarantee because behaviors are usually uncontrollable and variable. In view of the above deficiencies, this paper proposes a semantic retrieval method based on user intention recognition. Firstly, the paper introduces a glossary model to indicate different facets of relations of same type from different dimensions, and mine the inner association among keywords deeply. Secondly, by analyzing the structural relationship of multiple keywords, it acquires the user’s intention which is represented by four parameters using retrieval intention representation method, and gives a specific word expansion strategy directed towards different intentions. Finally, the effectiveness of the method is verified by an example which proves that the method can obtain more knowledge matching with the retrieval intent.