Abstract
With regard to the drastic development of information technologies, the interaction channels between enterprises and customers are more and more complicated. Many enterprises have realized that electronic CRM mechanisms can provide efficient and cost-effective customer services. In addition to organize the Frequently Asked Questions (FAQs) of customers, enterprises should gradually extract amd maintain the customer service knowledge in the electronic contact centers. Considering the operation cost and service quality of contact centers, the traditional service mechanism for cutomer queries should be re-engineered to reduced the human efforts required for question answering. This paper proposes an integrated model with automatic question-answering and knowledge summarization mechanisms. The integrated model consists of six main phases including construction of FAQs and keywords, question correlation analysis, meaning unit (MU) analysis, information intensity analysis, MU correlation analysis and MU linkage. The basic idea of the model is to explore the correlation between the target question and FAQs based on the keyword distribution. Based on the derived correlation, the candidate answers of the target question can be determined. Furthermore, according to the information intensity of MUs in the candidate answers, critical MUs are extracted and linked as the final answer to the target question. By applying of the automatic question-answering mechaism in the contact centers, interaction relationship between customers and contact centers can be improved and long-term customer relationship can be guaranteed. Moreover, the proposed approach can also be incorporated into the e-training systems for efficient query-replying and knowledge management systems for accurate information retrieval.