- 學位論文
智慧型系統於應用資訊系統使用者問題分類之研究
The Study of Using Intelligent System to Classify Applied Information System End-user Queries and Problems
摘要
隨著台灣產業外移到人工成本較低的中國以及其他東南亞國家，為了地區性以及即時性，應用資訊系統之維護與服務也隨之訓練由當地員工來做服務窗口，在客服支援服務實行多年以來，針對當地員工流動率非常高且無法在短時間內訓練新進人員熟悉資訊系統的狀況下，如何讓還不了解資訊系統的客服中心資訊人員，快速地了解使用者的問題，處理使用者的問題，將是目前首要需解決的問題。 本研究嘗試以類神經網路技術，針對以往的客服紀錄做分析研究，以決策樹找出具有影響的因子，並依此因子訓練倒傳遞網路，以做為新問題分類辨識的模擬結果。本項研究除提出模擬結果外，也詳述分類不準確的原因。本項研究成果可提供未來客服中心資訊人員在處理使用者問題時做為初步問題分類的參考，並協助客服人員使用自我協助方式解決問題
並列摘要
Taiwanese enterprises were forced to relocate their manufacturing bases to regions where provide cheaper labor such as the neighboring areas in China and other Southeast Asian countries. To improve resource utilization and to cut service-resolution times, we have to train local area employees to maintain information system and train them to be call center contact window. According to the high departee rate of employees, it is very difficult to train new member to realize the whole system within a short time. So, how to let call center IT members promptly understand the nature and severity of end-users queries become the top problem to be solved. In this thesis, the training data were collected from a company in 2007~2008. By utilizing the learning capability of back-propagation neural networks, we can predict the cause of the reason. We also apply the decision tree method to extracting the more influential factor for classification. The extracted factors are then become the inputs of back-propagation network to classify end-user queries and problems. Not only the simulation results are provided, but also a detailed discussion about the false prediction is given. The present work can be used as a reference for analyzing the reason of end-user queries and help call center members to solve the problem by self service.