Bandit Algorithms in Recommender Systems
by Dorota Glowacka (Helsinki University)
The multi-armed bandit problem models an agent that simultaneously attempts to acquire new knowledge (exploration) and optimize his decisions based on existing knowledge (exploitation). The agent attempts to balance these competing tasks in order to maximize his total value over the period of time considered. There are many practical applications of the bandit model, such as clinical trials, adaptive routing or portfolio design. Over the last decade there has been an increased interest in developing bandit algorithms for specific problems in recommender systems, such as news and ad recommendation, the cold start problem in recommendation, personalization, collaborative filtering with bandits, or combining social networks with bandits to improve product recommendation. The aim of this tutorial is to provide an overview of the various applications of bandit algorithms in recommendation.
This introductory 90-minute tutorial is aimed at an audience with some background in computer science, information retrieval or recommender system who have a general interest in the application of machine learning techniques in recommender systems.
Date
Thursday, Sept 19, 2019, 09:00-10:30
Location
Auditorium
Concept to Code: Deep Learning for Multi-task Recommendation
by Omprakash Sonie (Flipkart)
Deep Learning has shown significant results in Computer Vision, Natural Language Processing, Speech and recommender systems. Promising techniques include Embedding, Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN) and its variant Long Short-Term Memory (LSTM and Bi-directional LSTMs), Attention, Autoencoders, Generative Adversarial Networks (GAN) and Bidirectional Encoder Representations from Transformer (BERT).
Multi-task learning (MTL) has led to successes in many applications of machine learning. We are proposing a tutorial for applying MTL for recommendation, improving recommendation and providing explanation. We cover few recent and diverse techniques which will be used for hands-on session.
We believe that a self-contained tutorial giving good conceptual understanding of MTL technique with sufficient mathematical background along with actual code will be of immense help to RecSys participants.
Date
Thursday, Sept 19, 2019, 11:00-12:30
Location
Room 104+105
Material
Fairness and Discrimination in Recommendation and Retrieval
by Michael Ekstrand (Boise State University), Robin Burke (University of Colorado) and Fernando Diaz (Microsoft Research Montreal)
Fairness and related concerns have become of increasing importance in a variety of AI and machine learning contexts. They are also highly relevant to recommender systems and related fields such as information retrieval; however, translating algorithmic fairness constructs from classification, scoring, and even many ranking settings recommendation and other information discovery settings is not a straightforward task. This tutorial will help to orient RecSys researchers to algorithmic fairness, discuss how fairness concepts do and do not translate from other settings such as supervised machine learning, and provide an introduction to the growing literature on this topic.
Date
Thursday, Sept 19, 2019, 09:00-12:30
Location
Room 102+103
Material
Multi-Stakeholder Recommendations: Case Studies, Methods and Challenges
by Yong Zheng (Illinois Institute of Technology)
Recommender systems are able to produce a list of recommended items tailored to user preferences, while the end user is the only stakeholder in these traditional systems. However, there could be multiple stakeholders in several real-world applications or domains, e.g., e-commerce, advertising, educations, dating, job seeking, and so forth. Take car advertising for example, kids are one of the end users who watch Ads, but they may not be the target audience or potential buyers from the perspective of car producers. As a result, we believe that recommendations are necessary to be produced and delivered by balancing the needs of different stakeholders. This tutorial covers the introductions to multi-stakeholder recommender systems (MSRS), introduces multiple case studies, discusses the corresponding methods and challenges to develop MSRS. Particularly, a demo based on the MOEA framework will be given in the talk by using a speed-dating dataset.
Date
Thursday, Sept 19, 2019, 09:00-10:30
Location
Room 104+105
Recommendations in a Marketplace
by Rishabh Mehrotra (Spotify Research) and Benjamin Carterette (Spotify Research, University of Delaware)
nline marketplaces have witnessed an explosive growth by facilitating efficient interactions between multiple stakeholders, including e.g. buyers and retailers (Amazon), guests and hosts (AirBnb), riders and drivers (Uber), and listeners and artists (Spotify). A large number of such marketplaces rely on machine learning powered matching engines connecting consumers with suppliers by acting as a central platform, thereby finding the right fit and efficiently mediating economic transactions between the two sides.
In this tutorial, we consider a number of research problems which need to be address when developing a recommendation framework powering a multi-stakeholder marketplace. We begin by contrasting traditional recommendations systems with those needed for marketplaces, and identify four key research areas which need to be addressed. First, we highlight the importance of a multi-objective ranking/recommendation module which jointly optimizes the different objectives of stakeholders while serving recommendations. Second, we discuss different ways in which stakeholders specify their objectives, and highlight key issues faced when quantifying such objectives. Third, we discuss user specific characteristics (e.g. user receptivity) which could be leveraged while jointly optimizing business metrics with user satisfaction metrics. Furthermore, we highlight important research questions to be addressed around evaluation of such systems. Finally, we end the tutorial by discussing various diferent case studies and highlight recent findings.
Material
https://rishabhmehrotra.github.io/recs-in-marketplace/
Date
Thursday, Sept 19, 2019, 09:00-12:30
Location
Room 202+203, 204+205
SMORe: Modularize Graph Embedding for Recommendation
by Chih-Ming Chen (National Chengchi University), Chuan-Ju Wang (Academia Sinica), Ting-Hsiang Wang (Academia Sinica) and Ming-Feng Tsai (National Chengchi University)
Respectively, graph facilitates fusing complex systems of interactions into a unified structure and distributed embedding enables efficient retrieval of entities. When combined, graph embedding captures relational information beyond entity interaction and towards a problem’s underlying structure. This session will start by brushing up on the basics about graphs and embedding methods and discussing their merits. We then dive into using the mathematical formulation of graph embedding to derive the modular framework: Sampler-Mapper-Optimizer for Recommendation, or SMORe. We demonstrate existing models used for recommendation, such as MF and BPR, can all be assembled using three basic components. The tutorial is accompanied by a hands-on session, where we show how graph embedding can model complex systems through the multi-task learning and the cross-platform data sparsity alleviation tasks.
The 90-minute tutorial targets audiences who are interested in an overview of graph embedding in recommendation, speeding up model development, or using the toolkit’s preloaded models for applications.
A basic familiarity in collaborative filtering, recommendation models, embedding methods, and a rudimentary understanding about graphs is recommended but not required, as the basics will be covered. All materials, including slides, hands-on code, and the open-source project, will be publicly available after the tutorial.
Date
Thursday, Sept 19, 2019, 11:00-12:30
Location
Auditorium
