Chapter V: Rule-Based Parsing for Web Data Extraction

David Camacho, Universidad Carlos III de Madrid, Spain
Ricardo Aler, Universidad Carlos III de Madrid, Spain
Juan Cuadrado, Universidad Carlos III de Madrid, Spain

OVERVIEW

How to build intelligent robust applications that work with the information stored in the Web is a difficult problem for several reasons which arise from the essential nature of the Web: the information is highly distributed, it is dynamic (both in content and format), it is not usually correctly structured, and the web sources will be unreachable at some times. To build robust and adaptable web systems, it is necessary to provide a standard representation for the information (i.e., using languages such as XML and ontologies to represent the semantics of the stored knowledge). However, this is actually a research field and usually most web sources do not provide their information in a structured way.

This chapter analyzes a new approach that allows us to build robust and adaptable web systems by using a multi-agent approach. Several problems, including how to retrieve, extract, and manage the stored information from web sources, are analyzed from an agent perspective. Two difficult problems will be addressed in this chapter: designing a general architecture to deal with the problem of managing web information sources; and how these agents could work semiautomatically, adapting their behaviors to the dynamic conditions of the electronic sources.

To achieve the first goal, a generic web-based multi-agent system (MAS) will be proposed, and will be applied in a specific problem to retrieve and manage information from electronic newspapers. To partially solve the problem of retrieving and extracting web information, a semiautomatic web parser will be designed and deployed like a reusable software component. This parser uses two sets of rules to adapt the behavior of the web agent to possible changes in the web sources. The first one is used to define the knowledge to be extracted from the HTML pages; the second one represents the final structure to store the retrieved knowledge. Using this parser, a specific web-based multi-agent system will be implemented.