Personalized and preference-based recommender system w/prob relational models

A personalized and preference-based recommender system using probabilistic relational models

Résumé

Recommender systems aim at discovering potentially interesting items for users from a (usually large) collection of items. Most of the recommender systems implemented today deal with the problem of recommending simple, inexpensive, or frequently purchased items, such as books, movies, music, news, restaurants etc. and usually demand users’ profile for making recommendations. Relatively few works on the application of recommender systems for recommending expensive or less frequently purchased items (e.g., real estate properties, flights etc.) can be found. This talk will present a recommendation approach that uses probabilistic relational models for making personalized recommendations from users’ preferences in a system where items are not purchased frequently and user profiles are not readily available. Related Meetups:
Apprentissage de réseaux bayésiens à partir de bases de données de graphes (https://www.meetup.com/Nantes-Machine-Learning-Meetup/events/225705900/)
Les réseaux bayésiens (https://www.meetup.com/Nantes-Machine-Learning-Meetup/events/221520093/)

Bio

Rajani Chulyadyo has a Bachelor’s degree in Computer Engineering from Tribhuvan University, Nepal. After working as a Software Engineer in D2Hawkeye Services Pvt. Ltd. (currently known as Verscend Technologies Pvt. Ltd.) for 3 years, she did Erasmus Mundus Master Course in Data Mining and Knowledge Management in the University of Nantes and the University of Bucharest. She received the PhD degree in Computer Science from the University of Nantes in 2016. During her PhD, she was also involved in DataForPeople, Nantes as a researcher. Her domain of research includes probabilistic relational models, recommender systems and spatial data. Since 2013, she has been contributing to the development of PILGRIM, a software tool for working with probabilistic graphical models.

Reading

Chulyadyo, Rajani. A new horizon for the recommendation: Integration of spatial dimensions to aid decision making. PhD Thesis. Université de Nantes, 2016. URL (https://tel.archives-ouvertes.fr/tel-01422348/document)

Chulyadyo, Rajani, and Philippe Leray. "A personalized recommender system from probabilistic relational model and users’ preferences." Procedia Computer Science 35 (2014): 1063-1072. URL (http://www.sciencedirect.com/science/article/pii/S1877050914011582)

Getoor, Lise, et al. "Probabilistic Relational Models." STATISTICAL RELATIONAL LEARNING (2007), Chapter 5: 129. URL (https://www.cs.umd.edu/class/spring2008/cmsc828g/Papers/srlbook-ch5.pdf)

Language

The talk will be in English.