A Study of Integrating AHP and Case–based Reasoning into Case epresentation Systems– an Empirical Study from the Bureau of Investigation in Taiwan

• Main Authors: Chiang,Shu-Mei, 姜淑美
• Other Authors: 李俊民
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/69529898961377743915

碩士 === 輔仁大學 === 資訊管理學系 === 93 === In recent years, the Bureau of Investigation, Ministry of Justice (MJIB) continuously attempts revising the new organic law, explicitly building its standard responsibility, changing people’s mind of mystique, and actively investigating the major domestic crime, for current important obligations and establishing a good reputation of MJIB. Under the technological development of decision science and information systems, the computer center of MJIB has adopted and applied information systems for many years, and already become an organization of modernization and information turns. Therefore, the main purpose of this study is to apply the information science and technology to speed up the progress of criminal investigation, help collect the criminal data and information, and even develop a knowledge management support system based on the experience of case investigation, so as to enhance people’s confidence for MJIB investigating any criminal case. On the other hand, unstructured query languages are an important element for building a case-based reasoning (CBR) system and fuzzy query theories are mostly used as the foundation of the searching method of CBR. Specifically, the weight calculation in similarity model used for acquiring cases in database is usually neglected or affirmed by the decision maker oneself, as a result the cases obtained by comparing the similarity measures cannot completely satisfy the requirements of users. Aims at the problem described earlier, not only analytic hierarchy process (AHP), fuzzy query and professional knowledge are used for obtaining various characteristics, priorities and their weights of each case from the related class set stored in MJIB’s database, but also a prototype system of new and improved CBR model is built after integrating the results derived from AHP, so as to support decision making of case investigation and help the detective probe each case. Besides, the follow-up detective of MJIB can absorb experience and knowledge obtained from past cases, and can fast get at the key point of each case, in order to reach the efficiency of speeding up the rate of each case investigation, and understand the detective strategies and results of other similar criminal offence in the system. The empirical study uses the data in database of the case representation system of MJIB, and several research domains such as “the corruption defends”, “the drug defends”, “the economy crime defends”, “money laundering defends”, “international terrorism of anti-system”, etc., are identified and adopted in current CBR models, which only use fuzzy query in their similarity measures, but the improved CBR model integrates fuzzy query, AHP and professional knowledge into the similarity measure. Then, two models are tested for comparing their efficiencies and the rates of accuracy, respectively. Finally, different from previous researches, empirical results show that the improved CBR model has a statistically significant improvement. More empirical data of various types, incorporation of AHP, modification of similarity measure, or more empirical tests will be beneficial to the contribution of future research.