Abstract: The network is the largest collection of automatically accessible documents, which makes the richest source of information in the world. The different people may have distinct queries to search results and retrieve their results when they submit query in a search engine. The problem of clustering the feedback sessions are addressed in the dissertation.
The dissertation ““WEB BASED INFORMATION RETRIEVAL USING DYNAMIC CLASSIFIED AVERAGE PRECISION CRAWLING APPROACH” provides the best precision value and accuracy to the search results. The novel approach is inherited here to infer user goals by analyzing search engine query logs. The titles, snippets are created based on the clicked sequence of the query. Combining the titles and snippet, the feature representation can be derived. Feedback sessions are constructed from click through and can efficiently reflect the information needs of the user. They adopt a novel approach to generate the pseudo documents to better represent the feedback sessions for clustering.
Considering that if user search goals are inferred properly, the search results can also be restructured properly, since restructuring web search results is one application of inferring user search goals. In previous work the “Classified Average Precision” to evaluate the restructure results and describes the method to select the best cluster number. Previous studies have mainly focused on using manual query based logging examination to identify user web search results. In this dissertation they show whether and how they can automate this goal-identification process.
Here a new criterion “Dynamic Classified Average Precision Crawling (DCAPC)” is proposed to evaluate the performance of inferring user search goals. The dissertation proposes DCAPC (Dynamic Classified Average Precision Crawling) approach, an evaluation method based on restructuring web search results to evaluate whether user search goals are inferred properly or not and give better precision values. Experimental results are presented using user click-through logs from a commercial search engine to validate the effectiveness of the proposed methods. The quality of the extracted trait proposed DCAPC query logs as a helpful, though little explored, resource for in sequence data extraction.
Title: Web Based Information Retrieval Using Dynamic Classified Average Precision Crawling Approach
Author: R.Geethanjali, Mr.S.Muruganantham
International Journal of Computer Science and Information Technology Research
ISSN 2348-120X (online), ISSN 2348-1196 (print)
Research Publish Journals
