| Summary: | 碩士 === 靜宜大學 === 資訊碩士在職專班 === 98 === Facing the highly competitive market environment, customer relationship management has become one of the key factors of the successful corporations. Nowadays, the internet is a necessary component for companies to do external and internal communications. Therefore, to build a speedy and accurate customer response website system is more and more important and can''t be ignored today.
This study is to classify customer''s questions by using data mining classification method on un-structural contents. The method here can find out the right customer service person to answer for various questions rapidly, so it can shorten the original processing time to reduce corporate running cost and increase the satisfaction and royalty of customers.
The research raw data consists of the customer queries sentences on the website of one Taiwanese ERP(Enterprise Resource Planning) software development company. First, the study uses the Chinese word segmentation system developed by Chinese Knowledge Information Processing Group, Institute of Information Science in Academia Sinica, to analyze the customer query sentences. Second, the study generates the customers'' query keyword list. Third, the study builds two customer''s question classification models by using the decision tree method. The experimental results show that the accuracies of two models are 92.22% and 85.28%, respectively. This can prove that above procedure are effective on analyzing customers'' various queries. Finally, this study implements an automatic classification prototype system for customer query contents. The prototype system here can be integrated into customer relationship management system in the future.
|
|---|
