Date of Award

Spring 5-4-2021

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Yubao Wu

Second Advisor

Xiaojun Cao

Abstract

Online Social Networks (OSNs) have become prevalent in people’s daily life. Facebook, Twitter, and Instagram are among the most popular social networking platforms. There are over 3.6 billion people using social networks worldwide in 2020. The number is still expected to grow with the development of portable smart devices. With a large number of users, the impact of online social networks increases incredibly. The social network is not only changing how we communicate and interact with each other in daily life, but also how people make decisions.

Social graphs or networks are often used to depict the personal relations of Internet users on these platforms. In a social network, the nodes could represent individuals in the network, and the edges could represent the social interactions between them. It is crucial to mine useful knowledge from the huge and complex social network graphs efficiently. One of the

fundamental problems in Online Social Network is measuring the similarity between nodes effectively and efficiently. It is the first step for many applications like querying and ranking, community detection and link prediction.

As a result, we propose a series of works for effectively retrieving meaningful information in Online Social Networks. Firstly, we investigate the similarity measurement and propose a similarity measurement model to help measure the proximity between nodes in a graph. The proposed approach is validated by real-world social networking data. Then, we investigate the credibility of social influencers and propose a framework for content and influencer trustworthiness in online social networks. The texts and images are jointly considered and properly balanced. The proposed framework works as an approach to deal with the challenges for trustworthiness in the current online social networking environment in the application level. Then, we investigate utilizing online social networks to gather open-source intelligence from the Coronavirus-related events. All the proposed solutions are thoroughly discussed and validated with extensive evaluations.

DOI

https://doi.org/10.57709/22565128

File Upload Confirmation

1

Download

Share

COinS