Improvement of web information retrieval using swarm intelligence

Dissertant

Hadi, Mustafa J.

Thesis advisor

Abd Allah, Hasanayn Samir

University

University of Technology

Faculty

-

Department

Computer Sciences Department

University Country

Iraq

Degree

Master

Degree Date

2013

English Abstract

Information retrieval (IR) is the science of retrieving a subset of documents that satisfy the user’s need from a large collection of documents.

Traditionally, the only ones who were interested in IR field are the information gatherers and catalogers such as librarians.

However, with the advent of the Web, information access on the Web is the main problem of the so called Web Information Retrieval (Web IR) and the problems in the field of IR have become interesting real-world problems for many institutions and organizations.

Web IR is the application of IR concepts to the Web and the Web search engine is an application of Web IR that makes Web searchable and easily accessible.

Despite the existence of substantial and sustained efforts to develop the performance of search engines, with the exponentially growth as well as the rapid change of the Web, the general Web search engines remain not able to find correct and timely information for users.

This thesis investigates the application of Swarm Intelligence (SI) principles to IR in order to improve the work of Web search engines.

As a result of the massive information on the Web, the classic query processing approaches are no longer able to respond to queries in real time, thus developing an alternative tool to address Web IR problem has become an essential and indispensable issue.

The thesis will show the significance and superiority of the use of the SI approaches to improve the performance of IR in the Web context by introducing of the so called SI-IR system.

SI-IR system is an abbreviation of an information retrieval system based on swarm intelligence approaches.

Two proposed models of SI-IR system namely MPSO_IR΄ and MABC_IR΄ are designed for Web IR.

MPSO_IR΄ and MABC_IR΄ refer to use of the Modified Particle swarm Optimization (MPSO) and Modified Artificial Bee Colony (MABC) algorithms respectively.

These two modified algorithms depend essentially on a new data structure called “Nearest-Lists” added to other traditional existing data structures to enhance the solution quality , this is the reason why the mark (΄) is put at the top of IR.

The thesis will show that each one of the two proposed models can be an alternative to palliate the complexity issue in terms of response time while producing a solution quality is relatively convergent or even better.

Experimental tests have been conducted on two well-known CACM and NPL collections.

Both are different in size, CACM is small while NPL is relatively large.

Numerical results exhibit the superiority and the benefit gained from using the MPSO and MABC approaches instead of the classic approaches.

Main Subjects

Information Technology and Computer Science

Topics

• Information retrieval

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-418208