Unified framework based on algorithmic approaches for information retrieval over the web

Abstract

This research proposes a Unified Framework named Information Retrieval System over the Web (IRS_WEB). The proposed framework is based on the hybridization of various Machine Learning techniques of Indexing, Clustering and Ranking.A comparative analysis of tools, techniques and algorithms was performed to identify the most suitable algorithms for significant retrieval of data over the Web. The architecture plan of theproposed IRS_WEB drew its intention from a Three-Tier Architecture with the composition of Data Tier , Application Tier , and Presentation Tier . The Data Tier focused on the acquisition and pre-processing of the dataset in terms of data cleaning, tokenization and lemmatization for the removal of anomalies. The Presentation Tier dealt with the user interface and tunneled user interaction with the application. Further, the Application Tier dwelled upon the conceptual design of this research framework which started with Keyword-based Indexing to derive quality matches and range-based query operations efficiently. Subsequently, Synonym-based Clustering was applied to indexed documents to get the relevant results that satisfied the constraints of the user query. Finally, the Score-based Ranking was performed to evaluate the complex type of data related to specified criteria. The Proposed Unified Framework IRS_WEB was then executed and implemented using Programming Language Python, Heroku Platform and Open source database MongoDB. newlineIn order to extensively test, evaluate and validate the Unified IRS_WEB Frameworkbased on experimental setup of three research studieswas performed. Research Study 1 evaluated the Proposed IRS_WEB Framework based on Performance Metrics of Precision, Recall and F-measure. Research Study 2 conducted the qualitative evaluation of the Proposed IRS_WEB Framework using the acquired dataset. Research Study 3 tested and evaluated the Proposed IRS_WEB Framework with already existing Information Retrieval System based on standard Performance Metrics. newline