(ed.) Intelligent Agents for Data Mining and Information Retrieval

The main contributions of this chapter can be summarized as follows :

• Our approach tries to obtain a reusable and portable software that could be used by different web agents to extract knowledge from specific web sources.

• The utilization of rules provides two advantages: the flexible modification of the parser behavior when the source changes; and easy reutilization of the well- tested rules for similar web sources.

• Once the API of the WebParser is analyzed by the programmer, it is easy to use it as a new skill module inside the web agent. This could improve the implementation of web-based multi-agent systems and gathering systems.

Currently, we have implemented an initial version of the WebParser, and have integrated it into several web agents that belong to a simple multi-agent web system.

However, there are several important points that will be addressed to obtain a fully portable and reusable software for extracting web knowledge. These points can be summarized as:

• To study the flexibility of the rules that can be defined in the WebParser. Is it possible to extract other types of stored knowledge with this simple representation?

• To study other agent-based and multi-agent technologies and frameworks that actually have been used by different researchers and companies, including ZEUS (Collis et al., 1998) and JATLite (Petrie, 1996), and to see if it is possible to implement web or wrapper agents which integrate the parser inside the agents implemented with those technologies.