SEARCHING WEB THROUGH QUERY ROUTING PLANS

In future web will not be a collection of documents but can become sets of linked data sources. Keyword search is a way to search this kind of web as in present scenario, but will impose a high cost. This is because keyword search has to be applied to every source to find a match for any of keywords of the query. Hence routing keywords only to relevant sources that can reduce the unnecessary cost of running keyword search queries over all sources is needed. Also in the linked web setting a query is better answered with the help of data present across multiple sources i.e. the result of search is relevant combination of sources rather than relevant source. This can be achieved through a novel method which we propose in this paper. This method involves generation of top-k routing plans which can direct the search to appropriate sources based on the presence of keywords .A graph representing relationships between keywords and the data elements related to them at different levels i.e. at level of keywords, data elements, element sets, and sub graphs that connect these elements is generated. This is called multi-level keyword-element relationship (KER) graph. A scoring mechanism is proposed for computing the relevance of routing plans by ranking elements as well as relationships thus providing best results to the query.