Date of Award

2010

Publication Type

Master Thesis

Degree Name

M.Sc.

Department

Computer Science

First Advisor

Christie Ezeife

Keywords

Applied sciences

Rights

info:eu-repo/semantics/openAccess

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Abstract

Web content data are heterogeneous in nature; usually composed of different types of contents and data structure. Thus, extraction and mining of web content data is a challenging branch of data mining. Traditional web content extraction and mining techniques are classified into three categories: programming language based wrappers, wrapper (data extraction program) induction techniques, and automatic wrapper generation techniques. First category constructs data extraction system by providing some specialized pattern specification languages, second category is a supervised learning, which learns data extraction rules and third category is automatic extraction process. All these data extraction techniques rely on web document presentation structures, which need complicated matching and tree alignment algorithms, routine maintenance, hard to unify for vast variety of websites and fail to catch heterogeneous data together. To catch more diversity of web documents, a feasible implementation of an automatic data extraction technique based on object oriented data model technique, 00Web, had been proposed in Annoni and Ezeife (2009).

This thesis implements, materializes and extends the structured automatic data extraction technique. We developed a system (called WebOMiner) for extraction and mining of structured web contents based on object-oriented data model. Thesis extends the extraction algorithms proposed by Annoni and Ezeife (2009) and develops an automata based automatic wrapper generation algorithm for extraction and mining of structured web content data. Our algorithm identifies data blocks from flat array data structure and generates Non-Deterministic Finite Automata (NFA) pattern for different types of content data for extraction. Objective of this thesis is to extract and mine heterogeneous web content and relieve the hard effort of matching, tree alignment and routine maintenance. Experimental results show that our system is highly effective and it performs the mining task with 100% precision and 96.22% recall value.

Share

COinS