Arabic Spam Detection in Twitter

Conference Paper

2016

Published in: Arabic Spam Detection in Twitter

Al-Humoud, Nora Al-Twairesh, Mawaheb Al-Tuwaijri, Afnan Al-Moammar and Sarah . 2016

Spam in Twitter has emerged due to the proliferation of this social network among users worldwide coupled with the ease of creating content. Having different characteristics than Web or mail spam, Twitter spam detection approaches have become a new research problem. This study aims to analyse the content of Saudi tweets to detect spam by developing both a rule-based approach that exploits a spam lexicon extracted from the tweets and a supervised learning approach that utilizes statistical methods based on the bag of words model and several features. The focus is on spam in trending hashtags in the Saudi Twittersphere since most of the spam in Saudi tweets is found in hashtags. The features used were identified through empirical analysis then applied in the classification approaches developed. Both approaches showed comparable results in terms of performance measures reported reaching an average F-measure of 85% for the rule based approach and 91.6% for the supervised learning approach.

• Conference Location: Portorož (Slovenia)
• Conference Name: The 2nd Workshop on Arabic Corpora and Processing Tools
• Sponsoring Organization: European Language Resources Association (ELRA)