Efficient Extraction of Top-k Instances from Web

Abstract

Finding proper information from web pages is very difficult. Because we face problems such as most of available data contains unnecessary information such as some product advertisements, Facebook or twitter posts. One more problem is, obtained data is not in structured format. To overcome these problems, we introduce a system which mainly focuses on extracting exact information in top-k list format. List data is very eventful source to retrieving information. This paper work on information extraction from top-k web pages which contains top-k instances for open domain knowledge based. For example-"Top 10 IT companies in India". As compare to structured information from web, Top-k list data is cleaner and ranked. Top-k data has interesting semantics. We propose a system which gives direct top-k list when user enters a search query within minimum time. Extraction of top-k list depends on 1] Extracting web URLs and its titles 2] Removing dust from web URLs 3] Using extraction algorithm extract exact top-k list.