Research on a Model of Extracting Persons’ Information Based on Statistic Method and Conceptual Knowledge

Research on a Model of Extracting Persons’ Information Based on Statistic Method and Conceptual Knowledge* XiangFeng Wei a, Ning Jia a,b, Quan Zhang a, HanFen Zang c aInstitute of Acoustics, Chinese Academy of Sciences, China bGraduate University of Chinese Academy of Sciences, China cCaptical University of Economics and Business, China {wxf, zhq}@mail.ioa.ac.cn, gnin_aij@sina.com, zanghf@163.com Abstract. In order to extract some important information of a person from text, an extracting model was proposed. The person’s name is recognized based on the maximal entropy statistic model and the training corpus. The sentences surrounding the person’s name are analyzed according to the conceptual knowledge base. The three main elements of events, domain, situation and background, are also extracted from the sentences to construct the structure of events about the person. Keywords: Person’s Name Recognition; Hierarchical Network of Concepts; Main Elements of Events; Semantic Parsing; Information Extraction 1. Introduction In news reports, person is one of the most important elements. On the internet politicians, famous enterprisers, celebrities and hot persons are always the focus what people want to know. It would be very useful and valuable to compile and extract the information about a person with the assistance of using automatic information extraction technology. ACE (Automatic Content Extraction) program is organized by NIST (National Institute of Standards and Technology, U.S.A). It is aimed to develop automatic content extraction technology to support automatic processing of human language in text form, including the detection and recognition of entity, value, time, relation and event. In the description of events, ACE defined the allowable roles of an event, such as person, place, position, etc. Persou, which is studied by Liu and etc. (2006), is a web search engine for searching persons’ information. It can distinguish automatically or semi-automatically the persons who own the same name, and produce their resume and activities. A fame evaluation system studied by Zan and etc. (2003) can calculate the correlative degree between the basic information such as name, specialty, affiliation, character words of a person and the information of a web page. If the count of the strong correlative web pages is greater than a threshold, the relative person will be a famous person. In above studies, they all adopted statistic method. The advantage of using statistic models is high efficient performance * This work is supported by the National Basic Research Program of China (973 Program, the contract No. 2004CB318104) and the Fund for Excellence by the Director of the Institute of Acoustics, Chinese Academy of Sciences (the contract No. GS13SJJ04). Copyright 2007 by XiangFeng Wei, Ning Jia, Quan Zhang, and HanFen Zang. 508 to process large-scale data, but the precision is restricted by training set and it is not suitable to process sparse individual data, although the data is very important to the user. This paper proposed an model of information extraction based on statistic algorithm and conceptual semantic knowledge to extract persons’ information from the text. The primary information of a person includes name, gender, age, nationality, department in organization, …

Content truncated to 500 words