The Study of Applying Case-Based Reasoning to Explore Troubleshooting Diagnosis of Software Technical Problems

• Main Authors: SHIH FONG-MEI, 施鳳美
• Other Authors: Weng Sung-Shun
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/41147851618178042639

碩士 === 輔仁大學 === 資訊管理學系 === 94 === Currently the original company of the software faces the problems of immediate and professional maintenance job. Checking maintenance manual is inconvenient. Moreover, searching the internal FAQ website has the problem of returning too many or incorrect suggested solutions. They cannot help the software technical support analysts immediately reply customers’ usage or operational problems. Using the diagnosis platform will help customers quickly find out solution methods and increase their ability to self-solve their own problems. By so doing, it can save the cost of the on-line support for technical problems of the software company as well as increase its customers’ loyalty and satisfaction in the meantime. This is the main focus of this research. The purpose of this research is to build up a diagnosis platform for a set of software technical problems. The platform can record the characteristics of problems and the solution methods and keep them in the case database. They can be reused again. When a new case example happens, the platform can be used to retrieve the most alike case from the case database and provide the solution of the new case example. When it is necessary, it can revise the solution method of the case example or add the new case example. This research collects history data and gets the help from experts, using theories of case-based reasoning and its similarity formula as its basic concept. In the calculation of the similarity, for the attributes of category and version of operation system, it makes use of Oracle rule-based classification method to classify them, and compares whether old values and new values after classification are consistent or not. It also makes use of Oracle Fuzzy Matcher function to compare whether both values of problem’s description are similar or not. The subject of this research is focused on software technical problems. It bases the description of the problem’s characteristics to list the solution methods. The attributes and similarities between problem indexes along with the past experiences are used to solve new problems. We establish a FAQ case database for problems and maintenance along with a troubleshooting diagnosis system to assist users to find out the solution quickly. Moreover, users of this system can key in the characteristics of a problem, the system can find the similar solutions from the case database. Furthermore, users can modify the solutions of old cases at the right moment from the case database and insert new cases into the case database.