Back to search

Aristotle University Of Thessaloniki (AUTH)

Knowledge discovery from evolving social media data

Abstract

dc:description

Modern societies have been characterized as information societies, due to the unprecedented growth of information that has been witnessed during the last years, and its gradual exploitation for simplifying and improving everyday activities. The mass creation of information has been enabled primarily by the Web, and especially through the most recent rise of the Social Web. The latter encompasses the set of interactive Web applications (i.e. social media) that have been developed based on recent web technology innovations, drastically simplifying content creation for Web users, as well as the social network of users themselves. Social media data refer to information that can be sourced from social media applications, which can be related to: content (generated by users), activities (actions performed by users) and structure (associations between entities such as users). Given the heterogeneity, complexity, massive sizes and fast generation rates of such data, identifying effective processes and methodologies for discovering and leveraging knowledge is challenging. Social data mining addresses this challenge through the proposal of methods for harvesting knowledge based on content, activities, and / or structure. Challenges in social data mining impose the need for scalable, online methodologies that can process structure, content and activities in a time-aware approach. Such methodologies should additionally address the pervasive problem of fraud in social media, and also focus on deriving results that are useful for today’s societies, by enabling the interpretation of social data mining results in the context of real-world phenomena. Based on these remarks, three crucial problems have been addressed in this thesis, namely: (A) detection of communities (broadly considered as closely-knit groups of entities) in user and information networks; (B) discovery of fraudulent behavior in social media; (C) composite analysis of social media data for providing location-based services. This thesis contributes to addressing those problems through the development of effective methods and efficient frameworks for the discovery of knowledge based on social data mining. With respect to Problem A, the thesis advances approaches for the analysis and management of evolving user / information graphs (social structure mining). Methodologies for the discovery and tracking of communities in social media are proposed, as well as for their interpretation in relation to unfolding real-world events. Problem B is addressed through the proposal of methods for the discovery of fraudulent behavior in terms of the propagation of social media content, i.e. the specific case of the Rebroadcast fraud problem (social activity mining). Several distinctive patterns of fraudulent activity are identified and presented, along with a proposed methodology for the automatic detection of Rebroadcast fraud. Finally, in relation to Problem C, the thesis proposes and presents in detail frameworks for the analysis of shared content and activities in social media to derive location-based insights and leverage them for service provision (social activity and content mining).

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2015

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Giatsoglou, Maria
  • Γιάτσογλου, Μαρία

Subjects

dc:subject × 19

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/37542

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Giatsoglou, Maria; Γιάτσογλου, Μαρία. Knowledge discovery from evolving social media data. Aristotle University Of Thessaloniki (AUTH), 2015. http://hdl.handle.net/10442/hedi/37542