Extracting relevant information for a domain-specific search service using knowledge-based mining techniques

Opposite to general search engines, specialized search engines have the advantage to exploit specified properties of a special domain in order to allow a better search. For that, special methods are developed for the analysis of web sites and the retrieval is adapted. The example of a weather search service in Germany illustrates a knowledge-based, specialized search engine in this paper. There are a couple of WWW portals about weather information, climate, country sayings and health weather in Germany. The weather information providers present their data in a special way. These specialties are used for a knowledge-based data analysis of the web sites. Therefore an ontology and special analysis tools were developed. The goal is to present the most relevant information to the user in a web search service. To that, a three-stage process will be presented. In the first stage a traditional search process is started. In the second stage an expansion, an arrangement, and a classification of the document set is done using link tracing and ranking algorithms. And in the final stage the ontology concepts and relations are searched. The documents can be stored as URL-objects, HTML-sequences or java access classes. The data for the relevant information of an inquiry are aggregated from different resources using knowledge-based techniques.