Information Diffusion and Summarization in Social Networks

Kumar, Nagendra and Singh, Manish (2019) Information Diffusion and Summarization in Social Networks. PhD thesis, Indian institute of technology Hyderabad.


Download (6MB) | Preview


Social networks are web-based services that allow users to connect and share information. Due to the huge size of social network graph and the plethora of generated content, it is difficult to diffuse and summarize the social media content. This thesis thus addresses the problems of information diffusion and information summarization in social networks. Information diffusion is a process by which information about new opinions, behaviors, conventions, practices, and technologies flow from person-to-person through a social network. Studies on information diffusion primarily focus on how information diffuses in networks and how to enhance information diffusion. Our aim is to enhance the information diffusion in social networks. Many factors affect information diffusion, such as network connectivity, location, posting timestamp, post content, etc. In this thesis, we analyze the effect of three of the most important factors of information diffusion, namely network connectivity, posting time and post content. We first study the network factor to enhance the information diffusion, and later analyze how time and content factors can diffuse the information to a large number of users. Network connectivity of a user determines his ability to disseminate information. A well-connected authoritative user can disseminate information to a more wider audience compared to an ordinary user. We present a novel algorithm to find topicsensitive authorities in social networks. We use the topic-specific authoritative position of the users to promote a given topic through word-of-mouth (WoM) marketing. Next, the lifetime of social media content is very short, which is typically a few hours. If post content is posted at the time when the targeted audience are not online or are not interested in interacting with the content, the content will not receive high audience reaction. We look at the problem of finding the best posting time(s) to get high information diffusion. Further, the type of social media content determines the amount of audience interaction, it gets in social media. Users react differently to different types of content. If a post is related to a topic that is more arousing or debatable, then it tends to get more comments. We propose a novel method to identify whether a post has high arousal content or not. Furthermore, the sentiment of post content is also an important factor to garner users’ attention in social media. Same information conveyed with different sentiments receives a different amount of audience reactions. We understand to what extent the sentiment policies employed in social media have been successful to catch users’ attention. Finally, we study the problem of information summarization in social networks. Social media services generate a huge volume of data every day, which is difficult to search or comprehend. Information summarization is a process of creating a concise readable summary of this huge volume of unstructured information. We present a novel method to summarize unstructured social media text by generating topics similar to manually created topics. We also show a comprehensive topical summary by grouping semantically related topics.

[error in script]
IITH Creators:
IITH CreatorsORCiD
Singh, Manish
Item Type: Thesis (PhD)
Uncontrolled Keywords: Social network analysis, Information diffusion, Information summarization, Text mining, Influential users, Data characterization
Subjects: Computer science
Divisions: Department of Computer Science & Engineering
Depositing User: Team Library
Date Deposited: 06 Aug 2019 06:21
Last Modified: 06 Aug 2019 06:21
Publisher URL:
Related URLs:

    Actions (login required)

    View Item View Item
    Statistics for RAIITH ePrint 5871 Statistics for this ePrint Item