Improved text mining methods to answer Chinese e-mails automatically

The rapid development of E-commerce makes it difficult to deal with large numbers of customer e-mails quickly and effectively for enterprises. In order to solve the problem, we can employ the method of answering customer e-mail automatically based on text classification techniques. Classified into some appropriate classifications, customer e-mails can be answered using previously defined reply templates which correspond to classifications. In this paper, according to the feature of Chinese customer e-mails, we mainly use improved classification technique based on concept extraction to raise reply accuracy. In the process of classification, we consider the impact of linguistic context into the concept extraction, and establish two different classification criterions (product criterion and demand criterion) to raise classification accuracy. Correspondingly, in the selection of reply template, we combine the result of product analysis with the result of demand analysis to offer customers the most appropriate reply information.