Integrating automatic concept mapping and text summarization to enhance similarity-based search for information retrieval

مقدم أطروحة جامعية

Abu Alim, Maryam Rizq Allah Muhammad

مشرف أطروحة جامعية

Kanan, Ghassan Jaddu

أعضاء اللجنة

al-Shaykh, Asim A. R.
al-Dabbas, Umar Ṣuhaib
Kanan, Raid Karim

الجامعة

الأكاديمية العربية للعلوم المالية و المصرفية

الكلية

كلية نظم و تكنولوجيا المعلومات

القسم الأكاديمي

قسم نظم المعلومات الحاسوبية

دولة الجامعة

الأردن

الدرجة العلمية

دكتوراه

تاريخ الدرجة العلمية

2011

الملخص الإنجليزي

As an essential part of academic research, researchers have to review the literature by gathering notes about what was written in research area.

But, this process can be a time-consuming and straining for researchers who find themselves overwhelmed with too much published research papers.

Accordingly, there is an actual need for an efficient revision tool that can assist researchers to make sense of large documents easily and quickly, organize information, and find necessary resources to accomplish their works.

However, reverse outlining is extremely used as an effective and an efficient text summarization technique for taking notes and revising research papers, while concept mapping is a useful tool for visual representation and organizing information.

In addition, concept mapping is considered as a process of meaning making that provide a rich structure of semantically meaningful propositions in this study, a new approach is proposed with a vision that integration of reverse outlining and concept mapping will be useful for automatic generation of powerful descriptive concept maps which are easier to understand and to follow with.

Recently, automatic generation of concept maps for large documents have been gaining a lot of interest, In fact, generating a concept map that truly reflect the overall picture of a given document has not reached to a mature stage today.

If just selecting the important concepts, a map consisting of these concepts is not appropriate in terms of cohesion structure.

This problem occurs when the interpretation of some concepts in the map is dependent on that of others which are not selected as important ones, or when the proper connective is not involved, despite all the needed concepts are found.

For large documents, to address this problem, the proposed approach starts with extracting an informative summary of a given document, then based on the basics of concept map generation, re-summarizing the resulted summary into an indicative outline consists of a set of headlines (separated segments), and finally, in order to achieve the main gain of the discourse based summarization (i.e.

the ability to capture the dependency relations between the text segments and spot the concepts rather than its word components) and at the same time to avoid the problems of the possible absence or ambiguity of the lexical cues, and the incompleteness of the underlying knowledge bases, the outline is converted as a whole into a concept map as a method to produce a summary at a discourse level using the in common concepts as a connecting points of the separated segments in the outline.

Furthermore, in this approach, term frequency, titles and the leading sentences approaches are exploited to support developing a usable summary.

The implementation of the algorithm has been evaluated with encouraging results.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• الحوسبة
• استرجاع المعلومات

عدد الصفحات

139

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-306735