EVALUATION OF THE PROPOSED TOPIC SPECIFIC SEARCH ENGINE IN GEOSTATISTICS
Branislav Bajat · 2011
The Internet is not used simply for general collecting of information, but also for obtaining the information which are essential for specialists in many scientific fields. This paper introduces a topic specific search engine (TSE) designed for geostatistics, the discipline which is mostly used in geoinformation and spatial modeling community. TSE is able to learn the desired topic using a machine learning approach, based on a small number of examples i.e. links that point to relevant pages. The solution for the search engine proposed in this work could better respond to the needs of a specific community in finding, collecting and organizing only domain specific information by eliminating the noise in Web pages. The quality of the searching process was evaluated from the experts’s point of view, comparing proposed search engine and Google ® in series of equivalent searching sessions. For that purpose three types of standard queries were defined: query with one keyword, phrase query consisting of two keywords inside the quotes and phrase with two keywords plus one additional keyword joined together with Boolean AND operator. A comparison between the proposed solution and standard searching procedure using Google ® shows that our search engine outperforms