اخبار و رویدادها

An Individual-Social Framework for Modeling Diffusion in Twitter

An Individual-Social Framework for Modeling Diffusion in Twitter


By the appearance of online social networks, individuals have found more chances and more facilities to publish content. However, as the volume of these contents grows, the importance of detecting attractive ones increases. So, an important question is "what is an attractive content for public?".

In this project, this question is investigated by performing a comprehensive analysis of tweet content. In fact, the impacts of content-based features on the prediction of popularity and detection of topics are studied. The contents of tweets can be classified into individual and social contents. Individual contents are usually author's concerns about his/her events. Social contents discuss about topics and events that concern society. The main idea is that detecting individual and social contents can improve performance of popularity prediction and topic detection methods.

Thus, a dataset of Persian tweets was collected and the number of retweets of each tweet was considered as popularity measure. Then, an annotation process for accessing the semantic features of tweets was defined and the feature vector was formed by the lexical and semantic features.

The results of experiments indicate that social related features have more correlation with the popularity of tweets. Since retweeting is a social act, regarding this measure, social contents are more popular in society. Besides, detecting individual and social contents helped improve precision of topic detection methods.

Provider

محمد مهدوی
 

Supervisor

مسعود اسدپور
email: asadpour [AT] ut.ac.ir

 

 

ارائه دهنده پروژه

محمد مهدوی
عضو سابق
 

استاد راهنما

مسعود اسدپور
استادیار
اتاق: ساختمان جدید، 720
تلفن: 61114951
پست الکترونیکی: asadpour [AT] ut.ac.ir

 

 
آدرس کوتاه :