Abstract: In this project we combine the best ideas from the text extraction withthe help of character description and stroke configuration, web context searchand web mining with the help of semantic web and synaptic web at low entropy. First, we design a discriminative character descriptor. Second, we model characterstructure at each character class by designing stroke configuration maps.With the help of web context search, the extracted text is searched over thenet. In the proposed approach we have extended the context of user’s interestand developed an unsupervised algorithm to find the items of interest for theuser. Web mining is the application of data mining technique to automaticallydiscover and gathered information from web documents. It is used to find outthe relevant and efficient results from the web. Semantic-Synaptic web mining interlinks the web of data to different data sources at low entropy.