Seminar Link Prediction (Summer 2019)
|
|
Link prediction, or learning recommender systems, is a machine-learning task in which values assigned to pairs of nodes have to be predicted. Examples include predicting whether two members of an online social network are connected by an edge (i.e., "are friends") or predicting the value (or rating) that a potential customer assigns to a given product. Application scenarios for link prediction and large-scale empirical data abound due to ever-increasing use of online communication, collaboration, networking, purchasing, rating, or search. In this seminar we discuss, summarize, and compare seminal and recent work on link prediction and recommender systems. Participants can contribute in two ways: (1) by a theoretical contribution comprising a presentation and a term paper disussing published work and (2) by an experimental contribution comprising implementation, execution, and presentation of experiments for evaluating published link prediction methods on given empirical data. Topics will be presented in the first meeting (30 April 2019 at 13:30) and are also listed in the slides linked below. |
Schedule
| Introduction | Tue, April 30, 13:30-15:00 in L 602 |
| Presentations I | Tue, July 9, 13:30-16:45 in L 914 |
| Presentations II | Wed, July 10, 13:30-16:45 in PZ 901 |
Tentative schedule of participants' presentations
| No. | Topic | Date | Presenter | Slides |
|---|---|---|---|---|
| 1 | Graph similarity | July 9 | Dian Garkov | graphSimilarity-dian.garkov.pdf |
| 2 | Neighborhood-based methods | - | ||
| 3 | Neighborhood and factorization | - | ||
| 4 | Predicting edge signs | July 10 | Sahil Pasricha | LinkPredictionInSocialNetworks.pdf |
| 5 | Matrix factorization methods | - | ||
| 6 | Personalized Markov chains | July 9 | Guy Green | Personalized Markov chains- Guy Green.pptx |
| 7 | Factorization machines | July 9 | Markus Klatt | Link_Prediction_Klatt.odp |
| 8 | Predicting interaction events | - | Souvik Mondal |
Material
Most documents are only locally accessible - see possibilities for remote access.Slides
- slides.pdf (last updated: 30 April 2019)
Literature
Seminar topics
- (1) Liben-Nowell and Kleinberg (2007). The link-prediction problem for social networks. Journal of the American Society for Information Science and Technology, 58(7), 1019-1031. (local copy)
- (2) Bell and Koren (2007). Scalable collaborative filtering with jointly derived neighborhood interpolation weights. In Proceedings of the 7th Intl. Conference on Data Mining (pp. 43-52). IEEE. (local copy)
- (3) Koren (2008). Factorization meets the neighborhood: a multifaceted collaborative filtering model. In Proceedings of the 14th ACM SIGKDD (pp. 426-434). (local copy)
- (4) Leskovec, Huttenlocher, and Kleinberg (2010). Predicting positive and negative links in online social networks. In Proceedings of the 19th international conference on World wide web (pp. 641-650). ACM. (local copy)
- (5) Pilaszy, Zibriczky, and Tikk (2010). Fast ALS-based matrix factorization for explicit and implicit feedback datasets. In Proceedings of the fourth ACM conference on Recommender systems (pp. 71-78). ACM. (local copy)
- (6) Rendle, Freudenthaler, and Schmidt-Thieme (2010). Factorizing personalized Markov chains for next-basket recommendation. In Proceedings of the 19th international conference on World wide web (pp. 811-820). ACM. (local copy)
- (7) Rendle (2010). Factorization machines. In 2010 IEEE International Conference on Data Mining (pp. 995-1000). (local copy)
- (8) Lee, Nick, Brandes, and Cunningham (2013). Link prediction with social vector clocks. In Proceedings of the 19th ACM SIGKDD (pp. 784-792). (local copy)