Study On Law FrameNet Ontology-Based Semantic Retrieval

The traditional information retrieval systems are based on key words matching technique in the syntax layer. The system may miss some information that users really want and return some information that users don't want. It is the absence of general description of the information resources that makes the traditional system has no semantic processing capacity and information resources can not be shared totally. Essentially, Ontology is the description of the relations between objective existing concepts, so the Ontology-based information retrieval system, which formalized information resources and users inquiry with concepts of ontology, provides semantic retrieval which makes up for the flaw of the traditional keywords-based information retrieval system. It is better at recall and precision of the retrieval result.Though study on semantic retrieval has made great progress both in domestic and abroad. The present status is that study of theory is becoming more mature but study of application is backward comparably. Many mature methods of reasoning and retrieval are directly applied on ontology itself causing difficulty in information extraction and the expansibility and dynamical maintenance of the system is troubled. A mature system or a perfect, ontology-based semantic retrieval tool that can take effect in nature language on Website is still unavailable.Title of this paper is from the research projects "Study on the Construction of Chinese FrameNet Knowledge Ontology" aided financially by the State Social Science Fund. Based on the constructed Law FrameNet ontology, we explore the solution of semantic retrieval. By making frame semantic-based annotation on information resources and analyzing the user's inquiry sentence with frame semantic, we use the concept Graphs matching based method to measure the similarity between the retrieved information resources and the user's inquiry. Thus, both the precise matched information and that of semantic relative are retrieved; nevertheless the latter is mismatched with user's inquiry grammatically. This helps to improve the retrieval efficiency of law informationThe main contents of this paper are structured as follows. After giving some analysis to actuality of information retrieval in section 1 ,we point out the task of the study on Law Otology-based semantic retrieval in Section 1 and point out the significance of it. Section 2 analyzes the status and important role of ontology in semantic Web architecture and puts forward a ontology based information retrieval model and introduces its framework. It focuses on the functions and working flow of the suggested system and gives detailed description. Taking the Chinese law as the target domain, Section 3 goes deeply in the Law FrameNet Ontology based semantic retrieval system, which is divided into three subsystems: Law FrameNet Ontology management subsystem, Corpus management subsystem and Semantic retrieval subsystem. Section 4 makes a deeply research on the key technologies involved and its solution of this retrieval system, including understanding of questioning sentences and computation of sentence similarity. Section 5 is a trial semantic retrieval system; we use an example to simulate partial functions of the system in order to show the running principles. At the end, we draw conclusions and point to future work.