Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 46343 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Jalili, Mahdi
- Abstract:
- Regarding the analysis of real-world networks, the ability to detect regions of high density or “community” is of great importance in network science. The motivation behind this endeavor comes from many publications confirming that the agglomeration of relations is meaningful on its own. At the same time, there have been some attempts, although less than the unsigned route, toward the detection of community in groups of positive relations (similar to unsigned mode) that have, internally, as few negative ties as possible. As an example, signed Modularity is one of the few algorithms in this route.In this thesis, we first tried to analyze the structure of real networks to find out about the relative position of negative ties with respect to positive ones. The result of this analysis showed negative ties are naturally placed between dense positive ones. Moreover, in contrast to structural balance theorem, we demonstrated that triadic relation ++- is highly frequent in such networks,thus, could not be ignored.As the main goal, we succeeded to propose an algorithm for signed community detection,which out performs Modularity on artificial networks and, unlike Modularity, could be used to determine both the role and the position of negative ties in real networks. Apart from this, analyzing the formulation of proposed algorithm together with Modularity, we found a fundamental inconsistency in signed Modularity, which enlightens the route for future methods. Finally, utilizing the proposed method, we managed to determine the situation of negative ties between positive ones of specified density and, also, found out that negative ties are non-informative for two networks,but, they are informative for the other one, consistent with previous findings. However, in spite of previous methods, proposed tools give a more quantitative picture of negative ties from different scales of networks
- Keywords:
- Signed Networks ; Community Detection ; Structural Analysis ; Social Networks ; Structural Balance Theory ; Modularity
Digital Object List
محتواي کتاب
Bookmark