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: 52054 (01)
- University: Sharif University of Technology
- Department: Industrial Engineering
- Advisor(s): Haji, Babak
- Abstract:
- Due to the vast application of networks in the modeling of problems, the analysis of networks has been gaining a noticeable importance in recent researches. There are some cases in which we are looking for nodes that are highly connected to other nodes. However, the detailed information of the network structure is not always available; or because of complexity, it is not possible to investigate whole network and pick high-degree nodes; or investigating the network completely is too time-consuming. In such cases, algorithms which do not need the whole network information play a significant role. Friendship paradox as a topological feature of graphs can be helpful in designing such algorithms. It states that, "the number of your friends is less than the average number of the friends of your friends." In this research by means of analyzing the paradox holding probability for each node, we present a new algorithm based on friendship paradox. Afterwards, we compare our proposed algorithm against other existing algorithms. Furthermore, in special conditions we prove that our algorithm outperforms other algorithms. And then, in some examples, real network samples, and randomly generated graphs, we evaluate our algorithm performance. Afterwards, the factors effecting on our algorithm performance will be studied statistically. At last, we conclude that we can expect that our algorithm would perform better than other existing algorithms
- Keywords:
- Friendship Paradox ; Network Analysis ; Social Networks ; Graph High Degree Nodes ; Global Strategy ; Local Strategy ; Mutual Friends
Digital Object List
-
محتواي کتاب
- view
Bookmark