Design And Implemenation Of The Data Service Subsystem Of Financial Question Answering System

With the improvement of people’s living standard, management of personalmonery is paid more attention to. Hower, not everyone is financial expert, there arefew ways for them to get financial information conveniently.In this case, financialquestion answering system show a friendly way to get financial information througnatural language questions to individual investors.Data service system focused on the data processing of Financial QuestionAnswering System, and is composed of function modules like organization, storage,supply of semantic data, questions expansion, questions recommendation,fuzzy-search for unidentified tokens, organization of answers and so on.For the purpose of implementing of the organization, storage, supply ofsemantic data, this system put ontology concept and technology semantic web to use.This system construct domain top-level ontology through drawing standard modulediagram. Questions expansion is implemented based on the semantic data inontology base through searching technology in order to guid input of questioner. Inorder to recommend questions that may be needed by questioner, questionsrecommendation is implemented based on the historical sentences that can beanswered rightly and using technologies like search and string similar degreealgorithm that based on words density. In order to find semanteme of unknownwords, fuzzy-search is implemented based on recognized words in ontology basethrough processes like words spliting, searching and matching. Input of organizationof answer are operational orders that output by Questions Analytical System, and it’soutput is financial information in easily reading form that get through processes likedata extraction, mearging.Financial Question Answering System has already been released formally.Through the performance test, QPS of a single service is about10q/s. And throughthe user behavior analysis, ratio between count of sentences that input from inputbox and questions recommendation is about6:4. And ratio of users who are usingquestions expansion accounting for20%of the peoples who are using input box.