Semantic Search for Structured Web Finder

Semantic Web is one of the budding concepts in the data mining technology. Data mining involves extraction of data from a relevant and a huge source or a database or the World Wide Web. Whereas, semantic web involves cleaning, selection, transformation and presentation of data on the web. It covers wide knowledge from multiple disciplines like database technology, pattern recognition, visualization, machine learning and algorithm analysis. Further, it is responsible for fetching relevant data based on user's request. In the web world, there are many queries issued to search engines to find values from sorted data like train timing. In such scenarios, there is often a mismatch between the values of the user queries and the presented data on the web. Our proposed solution involves the usage of query logs both to generate entity synonym candidates, as well as to measure the likelihood of a candidate to be actual entity synonym. We also identify several clean up strategies to remove common and context-sensitive noise words. We further investigate the underlying cause of the formation of entity synonyms, by studying different entity synonym classes.