Structured web search in small domain

In this paper our aim is to bridging the gap between machine interpretation of search queries in search engines and actual user queries. Our primary concentration is to find a way to represent diverse search result pages in a structured way to reduce search query complexity and satisfy the need of diverse people. We first propose a traditional small scale search engine, called BuetSearch, for Bangladesh University of Engineering and Technology (BUET) domain, which has been designed by combining the PageRank algorithm and a proposed ranking and weighting scheme for Information Retrieval scoring. Then we propose a K-nearest neighbor algorithm based document classification algorithm for structured representation of search results. The results show that web search in small domain has merit over traditional large scale search engines and can be optimized to provide excellent results. With the application of structured approach, web search can be significantly simplified and the inconvenience of the users can be reduced.