綱路服務品質探勘與管制

一個受歡迎的網站可能會從網路會員處收到千百封的留言，最重要的訊息是屬於顧客關於技術需求及不滿意的怨言。本研究提出一個自動機制(WebQC)基於文件探勘和支援向量機(SVM)技術來分類顧客的留言，此機制可以自動過濾抱怨的留言並且正確地增加客服部的生產力以及顧客滿意度。本研究在抱怨率上使用了不良率管制圖(p-control chart)來檢查服務品質是否低於綱站運行的期望水準，並以一社群網站案例作為實驗案例，根據實驗結果顯示，其分類正確率或檢定力（如果留言為抱怨，SVH可以辨識為抱怨）的正確率超過83%（平均值為89%），而不良率管制圖可適時地反應出非隨機的狀況。

關鍵字

文件探勘；支援向量機；文件分類；補償不良率管制圖；網路服務品質

並列摘要

A popular Web site may receive hundreds of thousands of messages each day from site users. The moat useful customer feedback consists of technical requests and complaints. In this research, we propose the WebQC software to classify a user's feedback messages by using text-mining techniques and a support vector machine (SVM). Our software ran filter the messages as complaints or other kinds of messages automatically and thus effectively increase the productivity of the customer service department and as well as customer satisfaction. In this research, we employ the p-control chart on the complaint scale to check the quality of the Web site services if this quality is lower than expected. We used a community Web site as our test case. The empirical results show that our ability to classify a message accurately (i.e., if a message is a complaint, the SVM recognizes it as a complaint) is over 83% (on average 89%). Also, the p-control chart has the ability to reflect a normal situation in real-time.

並列關鍵字

Text mining ； Support vector machine SVM ； Classification ； compensating p-control chart ； Web service quality

參考文獻

吳志鴻(2001)。應用關鍵頁搜尋及知識分類技術於Q＆A系統之研究與設計(碩士論文)。元智大學資訊管理研究所。

Google Scholar

賴育昇(2002)。自然語言處理於網際網路常用答集檢索之研究(博士論文)。國立成功大學資訊工程學系。

Google Scholar

Berger, A.,Caruana, R.,Cohn, D.,Freitag, D.,Mittal, V.(2000).Bridging the lexical chasm: statistical approaches to answer-finding.Proceedings of the 23rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval.(Proceedings of the 23rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval).

Google Scholar

Burke, R.,Hammond, K.,Kulyukin, V.(1997).Natural language processing in the FAQ finder system: Result and prospects.1997 AAA1 Spring Symposium Technical Report.(1997 AAA1 Spring Symposium Technical Report).