Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection

Social networks and their usage in everyday life have caused an explosion in the amount of short electronic documents. Social networks, such as Twitter, are common mechanisms through which people can share information. The utilization of data that are available through social media for many applicat...

Full description

Saved in:
Bibliographic Details
Main Author: Alsmadi, Issa Mohammad Ibrahim
Format: Thesis
Language:English
Published: 2018
Subjects:
Online Access:http://eprints.usm.my/46679/1/short%20text%20classification%20using%20an%20enhanced%20term%20weghiting%20scheme%20and%20filter-wrapper%20feture%20selection24.pdf
http://eprints.usm.my/46679/
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Universiti Sains Malaysia
Language: English
id my.usm.eprints.46679
record_format eprints
spelling my.usm.eprints.46679 http://eprints.usm.my/46679/ Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection Alsmadi, Issa Mohammad Ibrahim QA75.5-76.95 Electronic computers. Computer science Social networks and their usage in everyday life have caused an explosion in the amount of short electronic documents. Social networks, such as Twitter, are common mechanisms through which people can share information. The utilization of data that are available through social media for many applications is gradually increasing. Redundancy and noise in short texts are common problems in social media and in different applications that use short text. However, the shortness and high sparsity of short text lead to poor classification performance. Employing a powerful short-text classification method significantly affects many applications in terms of efficiency enhancement. This research aims to investigate and develop solutions for feature discrimination and selection in short texts classification. For feature discrimination, we introduce a term weighting approach namely, simple supervised weight (SW), which considers the special nature of short text in terms of term strength and distribution. To address the drawbacks of using existing feature selection with short text, this thesis proposes a filter-wrapper feature selection approach. In the first stage, we propose an adaptive filter-based feature selection method that is derived from the odd ratio method, used in reducing the dimensionality of feature space. In the second stage, grey wolf optimization (GWO) algorithm, a new heuristic search algorithm, uses the SVM accuracy as a fitness function to find the optimal subset feature. 2018-12 Thesis NonPeerReviewed application/pdf en http://eprints.usm.my/46679/1/short%20text%20classification%20using%20an%20enhanced%20term%20weghiting%20scheme%20and%20filter-wrapper%20feture%20selection24.pdf Alsmadi, Issa Mohammad Ibrahim (2018) Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection. PhD thesis, Universiti Sains Malaysia.
institution Universiti Sains Malaysia
building Hamzah Sendut Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Sains Malaysia
content_source USM Institutional Repository
url_provider http://eprints.usm.my/
language English
topic QA75.5-76.95 Electronic computers. Computer science
spellingShingle QA75.5-76.95 Electronic computers. Computer science
Alsmadi, Issa Mohammad Ibrahim
Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
description Social networks and their usage in everyday life have caused an explosion in the amount of short electronic documents. Social networks, such as Twitter, are common mechanisms through which people can share information. The utilization of data that are available through social media for many applications is gradually increasing. Redundancy and noise in short texts are common problems in social media and in different applications that use short text. However, the shortness and high sparsity of short text lead to poor classification performance. Employing a powerful short-text classification method significantly affects many applications in terms of efficiency enhancement. This research aims to investigate and develop solutions for feature discrimination and selection in short texts classification. For feature discrimination, we introduce a term weighting approach namely, simple supervised weight (SW), which considers the special nature of short text in terms of term strength and distribution. To address the drawbacks of using existing feature selection with short text, this thesis proposes a filter-wrapper feature selection approach. In the first stage, we propose an adaptive filter-based feature selection method that is derived from the odd ratio method, used in reducing the dimensionality of feature space. In the second stage, grey wolf optimization (GWO) algorithm, a new heuristic search algorithm, uses the SVM accuracy as a fitness function to find the optimal subset feature.
format Thesis
author Alsmadi, Issa Mohammad Ibrahim
author_facet Alsmadi, Issa Mohammad Ibrahim
author_sort Alsmadi, Issa Mohammad Ibrahim
title Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
title_short Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
title_full Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
title_fullStr Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
title_full_unstemmed Short Text Classification Using An Enhanced Term Weighting Scheme And Filter-Wrapper Feature Selection
title_sort short text classification using an enhanced term weighting scheme and filter-wrapper feature selection
publishDate 2018
url http://eprints.usm.my/46679/1/short%20text%20classification%20using%20an%20enhanced%20term%20weghiting%20scheme%20and%20filter-wrapper%20feture%20selection24.pdf
http://eprints.usm.my/46679/
_version_ 1672611554018721792