Correlation Retrieval Of Criminal Transcript Information Based On Vector Space Model

Information retrieval can help users get to demand information fast and accurately, shorten the query time, improve work efficiency, and has been more and more people concern. A rapid and accurate method of text information retrieval for criminal record information query work is also necessary. Criminal record as one of the main carrier of investigation clue, will be a criminal case involved about criminal cases of the original information in text mode is presented to scout, prosecutors and judges. Criminal record by different observers, various cases of information is scattered in each copy of the record. The present approach uses artificial way for police statistics after each record information and retrieve the clues from the record or contradictory points, to other civilian police for further investigation. But there are repeated labor manual retrieval ways, so inefficient and subsequent investigation work need to retrieve the clues or contradictory points to determine the work direction and key. To retrieve the clues and the contradictory points work lag and subsequent investigation work will be stagnant, unable to meet the needs of the work. In view of this dilemma, the record information in urgent need of rapid retrieval methods can help police search out the required information, so as to solve the case soon.In this thesis, on the basis of criminal record instance, criminal case record in criminal cases the status of the chain of evidence and the difficulty in the process of the record in constructing the chain of evidence, the text information is obtained by the concept of text word association retrieval solution. Combined with the actual situation of the criminal record, we design a set of reasonable retrieval algorithm. In this thesis, the main work includes:1. Upon the current working method, we introduce the criminal case of taking a role in the construction of criminal evidence chain and construction difficulty.2. Aiming at these difficulties, in the victim record information for reference, we build record retrieval method based on vector space model, using cosine similarity measure mechanism, from the criminal suspect record text information to calculate the similarity of the record information.3. For record relevant retrieval algorithms based on the vector space model, we make experimental analysis. The experimental results show that the proposed algorithm has high accuracy and efficiency.