Research On Question Answering Oriented Financial Ontology Building Techniques

With the rapid development of Internet technology, massive information emerges,there is a growing need for fast, accurate access to information, but the traditionalinformation retrieval technology still has many deficiencies. Automatic questionanswering system (QA system) gives a more accurate, simple, intelligent way to provideusers with the information needed, which makes the research and application ofautomatic question answering get more and more attention.Most traditional QA systems are based on question and answer pairs knowledgebase, use the keyword indexing, this answering system answers low accuracy, andcannot meet the professional fields. Fundamentally, this is because such a knowledgebase answering system does not involve semantics. To solve this problem, this paperintroduces a domain-specific knowledge base construction method for QA system, themethod is based on the Semantic Web ontology technology, combined with the actualsituation in specific areas, finally, complete a domain question answering system basedon this knowledge base.This study content mainly includes three parts: ontology construction orientedfinancial sector, answering system construction based on the ontology knowledge base,and extracting Wikipedia knowledge to expand ontology knowledge. First, establish theinitial body structure of the financial ontology base on the actual situation, and crawl allcompanies profile information in Shanghai and Shenzhen exchange, filling the initialontology and get an initial knowledge base; due to the initial knowledge base containsonly generic and structured information on companies, and the knowledge baseconstruction is a continuous process of expansion and superposition, subjects attempt toexploit and extract useful knowledge from Wikipedia which is unstructured, to improvethe body structure and expand ontology coverage; finally, based on the ontologyknowledge base, build a financial QA system.This paper presents a semi-automatic ontology knowledge base constructionmethod: design initial body template, and then use domain knowledge constantlyreplenishing the body structure of ontology. And an unstructured text informationextraction method for ontology construction is proposed in the paper: use Wikipedia asthe knowledge source, exploit the infobox information, and apply machine learningalgorithms to extract information from unstructured Wikipedia pages.Based on the application actual requirements, this paper presents in detail a methodfor the design and implementation of ontology knowledge base QA system oriented.Experiments show that, the Wikipedia information extraction model of this paper canmeet the requirements of the ontology expanding, and the overall ontology construction method can be used effectively in the QA system.