Question Answering System For Specific Areas And Implementation On The NAO Mobile Robot Platform

Intelligent question answering system is different with the traditional search engine, user’s questions are more diversified, no fixed format of the form,natural language questions. Keyword based search engine returns to the user that may be some lists, but the intelligent question answering system do more directly and more intelligently, giving a clear answer. Therefore, compared with the search engine, question answering system is the more difficult. There are many ways to construction the question and answering system, some are based on the database, other are based the search engine. The database based question answering system needs to construct a huge knowledge base, which is a huge and tedious part of the workload. Question and answering system that based on search engine is a tedious process omits the knowledge base construction, mainly focused on the understanding of the question、answer extraction and ranking.With the development of information Webpage gradual, a lot of information can be obtained from the internet. At the same time, the teachers’ information of computer department of Nanjing University is not uniform, who want to understand the relevant information of the teachers without a standardized system to help them. The scattered and opaque information spend a lot of time to check. Based on this, this paper is from the perspective of practical demand, integrating all aspects of the information,to construct a question answering system based on the CS information of the teachers. This paper focuses on a restricted domain question answering system, which based on computer CS (Computer Science) teachers information to construct the corresponding knowledge base, including crawling content from different Web site entrance, Webpage structure analysising, making the further information extraction, data cleaning, data fusion, entity disambiguation, and producing structured knowledge base. Taking this so perfect CS knowledge base as the basis, at the same time,we take the group buying NAO humanoid robot Aldebaran company as the hardware platform to build intelligent question answering system iAskNao.In this paper, the main research is focused on constructing structurization knowledge base, the analysis and calculation of the query, and the knowledge base of Web visualization. The process of constructing the traditional Q & A system generally include:query analysis, information retrieval and answer extraction, the emphasis of this paper lies in the system building process, including the Web crawling, information extraction,data cleaning, construction of knowledge base, Web visualization and building a question answering system in the robot, which is to study and analyze the construction of intelligent question answering system. Finally, the whole procedure of the work are integrated in the NAO hardware platform. Nao robot comes with a speech recognition engine, but the drawback is that it is identity insert, thus limiting the question form, cannot satisfy users with diversed forms.Because of this, this paper adopts the cloud speech recognition method, the NAO robot through access to the user’s speech question, up to web cloud to identify, again to return to the character recognition,then do question analysis and answer retrieval and a series of processing. So customer service the NAO comes the problem of speech recognition, the question form is more diverse. For the latter part of the question and answer question in system development laid a very important foundation.