An intelligent interface agent for Web-based information retrieval

Abstract

Presents an intelligent interface agent, called DOLTRI-Agent (Distance and Open Learning Training Resource Information retrieval Agent), with three features: knowledge acquisition, autonomy and collaboration. It is applied to the World Wide Web-based information retrieval field. For the knowledge acquisition, there are currently three approaches: the end-user programming approach, the knowledge-based approach and the machine learning approach. Each has its advantages and disadvantages. Our interface agent adopts a hybrid of the latter two approaches, in order to combine their advantages and to counteract their disadvantages. Moreover, this interface agent extends the memory-based reasoning technique, adopting multiple user models and multiple analysis models. This intelligent interface agent has two types of autonomy. One is named concomitant autonomy, which includes information guidance and situation-specific help, and the other is called independent autonomy, which includes results pre-processing and information updating. Among the different users, collaboration is built in, in order to allow the knowledge owned by each user model to be shared.