Link Prediction in Inferred Social Networks
Thesis title in Czech: | Predikce spojení v odvozených sociálních sítích |
---|---|
Thesis title in English: | Link Prediction in Inferred Social Networks |
Key words: | odvozená sociální síť|predikce spojení|grafová data|grafová neuronová síť |
English key words: | inferred social network|link prediction|graph data|graph neural network |
Academic year of topic announcement: | 2020/2021 |
Thesis type: | diploma thesis |
Thesis language: | angličtina |
Department: | Department of Software Engineering (32-KSI) |
Supervisor: | doc. RNDr. Irena Holubová, Ph.D. |
Author: | hidden - assigned and confirmed by the Study Dept. |
Date of registration: | 17.10.2020 |
Date of assignment: | 19.10.2020 |
Confirmed by Study dept. on: | 14.12.2020 |
Date and time of defence: | 22.06.2021 09:00 |
Date of electronic submission: | 15.05.2021 |
Date of submission of printed version: | 21.05.2021 |
Date of proceeded defence: | 22.06.2021 |
Opponents: | Mgr. Ladislav Peška, Ph.D. |
Guidelines |
The knowledge of a social network of clients would bring various benefits to companies and businesses. However, an access to such data is highly limited. Recently there has occurred the idea of inferred social networks, i.e., networks that are not built by the people themselves, but inferred from the knowledge of their particular behaviour (e.g., usage of mobile phones, public transport, bank accounts etc.). This idea however brings many challenging problems.
The aim of this thesis is to focus on the problem of link prediction in an inferred social network using existing verified approaches. For this purpose the author will use real-world data from the financial sector and adapt the selected methods to the specific targets of this area. The result of the thesis will be an experimental exploration of selected suitable approaches for this new type of networks. |
References |
Holubova, I. - Svoboda, M. - Berhauer, D. - Skopal, T. - Pascenko, P.: Inferred Social Networks: A Case Study. BSMDMA@ICDM '19: Proceedings of the 2019 International Workshop on Big Social Media Data Management and Analysis, held in conjunction with ICDM '19, Beijing, China, November 2019.
Needham, M.: Link Prediction with Neo4j. 2019. https://medium.com/neo4j/link-prediction-with-neo4j-part-1-an-introduction-713aa779fd9 Liben-Nowell, D. - Kleinberg, J. The link prediction problem for social networks. In Proceedings of the twelfth international conference on Information and knowledge management (CIKM ’03). Association for Computing Machinery, New York, NY, USA, 556–559. 2003. DOI:https://doi.org/10.1145/956863.956972 Bringmann, B. - Berlingerio, M. - Bonchi, F. - Gionis, A.: Learning and Predicting the Evolution of Social Networks. Intelligent Systems, IEEE. 25. 26 - 35. 2010. 10.1109/MIS.2010.91. Kim, M. - Leskovec, J.: The Network Completion Problem: Inferring Missing Nodes and Edges in Networks. https://cs.stanford.edu/people/jure/pubs/kronEM-sdm11.pdf Zhang, M. - Chen, Y.: Link Prediction Based on Graph Neural Networks. NeurIPS 2018: 5171-5181 Bronstein, M.: Temporal Graph Networks. https://towardsdatascience.com/temporal-graph-networks-ab8f327f2efe |
Preliminary scope of work |
Možnost analyzovat sociální sítě klientů by se velmi hodila mnoha firmám, ale přístup k takovým datům je obvykle značně omezený. Myšlenka odvozených sociálních sítí spočívá ve vytvoření umělé sítě extrahované na základě znalostí chování klientů (např. používání mobilních telefonů, bankovních účtů, MHD apod.). Cílem práce je zaměřit se na problematiku predikce spojení v odvozené sociální síti. Vstupem práce bude síť vytvořená na základě reálných dat z oblasti bankovnictví. Autor v rámci práce vyzkouší a zhodnotí použitelnost vybraných metod predikce spojení pro tento specifický typ sociálních sítí a danou doménu. |