Epistemic Categorization for Analysis of Customer Complaints

摘要

We introduce the particular functionality of the Complaint Engine suite, the integrated complaint management component for mediating consumer disputes. We formulate the problem of epistemic categorization: whether a given complaint submitted by an upset customer is an adequate representation of company's product, service and attitude towards customers. To do that, a sequence of communicative actions and argumentation patterns in the course of complaint resolution (as described by a complainant) is analyzed. Instead of natural language processing of textual complaints we use two interactive forms: the first one, to specify communicative actions and argumentative links between their parameters, and the second one, to specify the argumentative relations between the major claims. We briefly outline the reasoning components involved in processing the above data to extract information which would then be expected to be truthful. We then perform the comparative evaluation of the involved reasoning units.