Equipped Search Results Using Machine Learning From Web Databases

Abstract

Database driven web pages play a vital role in multiple domains like online shopping, e- education systems, cloud computing etc. Such databases are accessible through HTML forms and user interfaces. They return the result pages which come from the underlying databases as per the nature of the user query. Such types of databases are termed as Web Databases. Web databases have been frequently employed to search the products online for retail industry. They can be private to a retailer/concern or publicly used by a number of retailers. Whenever the user queries these databases using keywords, most of the times the user will be deviated by the search results returned. The reason is no relevance exists between the keyword and Search Results. A typical web page returned from a Web Database has multiple Search Result Records. An easier way is to group the similar Search Result Records into one cluster in such a way the user can be more focused on his demand. It is proposed to develop a novel system called Clustering Search Results which extracts the data from the XML database and clusters them based on the similarity and finally assigns meaningful label for it. To do that the research work is aimed that, While the conventional searches simply extract the information and display it, the proposed system thoroughly analyzes it and clusters the similar results together and the dissimilar separated. The clustering activity is taken care not solely by the logics. It is shared with the xquery utility too. Thus the burden on clustering is v divided into two viz., xquery for partial clustering and bean component for cluster organization. Finally, the clustering turns into full-fledged functionality. The database for retrieving is made as eXtensible Markup Language (XML). It can consume large volume of data with smaller storage area; hence it is hierarchical in nature and semi structured. The most suitable label is assigned to every cluster created. It makes the search results meaningful in nature for better understanding. newline