Titre :	Author proling for social media users in Arabic language
Type de document :	texte imprimé
Auteurs :	Ayad, Rayane, Auteur ; Sadik Bessou, Directeur de thèse
Editeur :	Setif:UFA
Année de publication :	2021
Importance :	1 vol (50 f .)
Format :	29 cm
Langues :	Français (fre)
Catégories :	Thèses & Mémoires:Informatique
Mots-clés :	Author Proling Machine Learning
Index. décimale :	004 Informatique
Résumé :	Author Proling has been a popular research topic in the eld of NLP,and also the focus of a number of prominent shared tasks. Author Proling is an increasingly important issue, with applications in forensics, security, sales, and many more. In recent years, text available from social networks has become a primary source for author prole computational models, but current studies still focus largely on age and gender prediction, and in many cases are limited to the use of English text. Other languages and other author proling tasks remain somewhat less common. As a means of furthering this problem, , in this work we present initial results of a number of author proling tasks from a twitter corpus in the Arabic language. We will focus our work on all of the standard gender, age, and eld of work forecasting tasks but combine these features into a single category. For this purpose, we built a new data set in Arabic and explained the techniques we used in the preprocessing of the data set. In our study we applied NGrams combinations with both BOW and IDF-TF features to identify the best. During training, the use of IDF-TF with NGrams reached the highest performance overall. After we applied six machine learning algorithms, it was found that the logistic regression algorithm provides the best performance and its result
Côte titre :	MAI/0472
En ligne :	https://drive.google.com/file/d/1LQPllaYCzg2SJEhdBuwCLuxICLmhXKhu/view?usp=shari [...]
Format de la ressource électronique :	pdf

Author proling for social media users in Arabic language [texte imprimé] / Ayad, Rayane, Auteur ; Sadik Bessou, Directeur de thèse . - [S.l.] : Setif:UFA, 2021 . - 1 vol (50 f .) ; 29 cm.
Langues : Français (fre)

Catégories :	Thèses & Mémoires:Informatique
Mots-clés :	Author Proling Machine Learning
Index. décimale :	004 Informatique
Résumé :	Author Proling has been a popular research topic in the eld of NLP,and also the focus of a number of prominent shared tasks. Author Proling is an increasingly important issue, with applications in forensics, security, sales, and many more. In recent years, text available from social networks has become a primary source for author prole computational models, but current studies still focus largely on age and gender prediction, and in many cases are limited to the use of English text. Other languages and other author proling tasks remain somewhat less common. As a means of furthering this problem, , in this work we present initial results of a number of author proling tasks from a twitter corpus in the Arabic language. We will focus our work on all of the standard gender, age, and eld of work forecasting tasks but combine these features into a single category. For this purpose, we built a new data set in Arabic and explained the techniques we used in the preprocessing of the data set. In our study we applied NGrams combinations with both BOW and IDF-TF features to identify the best. During training, the use of IDF-TF with NGrams reached the highest performance overall. After we applied six machine learning algorithms, it was found that the logistic regression algorithm provides the best performance and its result
Côte titre :	MAI/0472
En ligne :	https://drive.google.com/file/d/1LQPllaYCzg2SJEhdBuwCLuxICLmhXKhu/view?usp=shari [...]
Format de la ressource électronique :	pdf