Efficiency of Unstructured Text Search Improving Methods in the Electronic Archive of Computer-aided Design Systems

摘要

This paper proposes an approach for improving the efficiency of the unstructured text documents search in the electronic archive of computer-aided design systems by increasing the pertinence and relevance. The authors propose a mathematical description of the presentation of information based on fuzzy sets, the context-aware search terms. There was developed an algorithm for constructing a search pattern of the document based on the proposed mathematical description. The researchers propose the method and the algorithm for clusters formation from associated search terms in the document that allows one to form a visual representation of the substantive content of the document as a graph. An algorithm for searching for semantically similar documents was developed and implemented. The proposed method of the semantically similar documents search automatization allows one to: clarify the information needs of users and build up a more detailed search query by introduction of a dialogue; take into account the context similarity of the of significant documents terms collection and of the term sample documents context. The conducted experiments demonstrate the effectiveness of the proposed approach.