Emergency Case Information Extraction And Retrieval Based On Knowledge Element

Emergency possesses sudden, complexity and derivatives with features of outstandi ng multidisciplinary and professional, so traditional pattern "predict-response" has been replaced by pattern "scenario-response". Emergency case is a true account of historica1incident scenario, which includes lots of information about objective things and their state change. Scientific and rational use of this information will have a positive role in support of emergency decision. However, currently vast majorities of emergency cas es are stored in files or database with free text or semi-structured text, which lack of depth combing. Case utilization is very low. Therefore, how to accurately extract val uable information from emergency case and effectively use this information is a probl em to solve.On basis of a systematic review of knowledge element theory, scenario theory an d current research status of information extraction and case retrieval, introduce commo n knowledge model into emergency field. According to emergency case content feature and scenario theory, reasonably reconstruct emergency case with event knowledge ele ment and object knowledge element, which is represented as background and foregrou nd based on knowledge element then stored into XML. In order to obtain information of objective things and their state change in emergency case, put forward a method f or emergency case information extraction based on knowledge element. Give build ste ps of case information extraction template that is customized and extraction rules. Esta blish information extraction process, then research algorithm of information extraction and missing information compensation. Moreover, for solving problems that include on e-sidedness and neglecting importance of environmental factors in previous case retriev al, according to case information extraction results, raise a method for emergency case retrieval based on knowledge element. Construct overall retrieval process and similarit y calculation process and research calculation methods of background similarity, foregr ound similarity and global similarity. For foreground similarity, propose concept of co mplete subscript positive sequence columns, then give steps to find out complete subs cript positive sequence columns from foreground fragiments similarity matrix. At last, t hrough applying proposed methods to mine accident case information extraction protot ype system and retrieval example verification, prove feasibility and efficiency. Research on methods for emergency case information extraction and retrieval base d on knowledge element will further promote realization of emergency decision-making model on basis of "scenario-response". On one hand, provide objective and true data support for research on emergency scenario analog simulation based on knowledge el ement. On other hand, lay theoretical foundation for realization of emergency case rea soning system based on knowledge element.