The RecSys 2010 Industry Panel Will Recommenders Kill Search? Recommender Systems – An Industry Perspective Ido Guy 1 , Alejandro Jaimes 2 , Pau Agulló 3 , Pat Moore 4 , Palash Nandy 5 , Chahab Nastar 6 , Henrik Schinzel 7 1 IBM Research, Israel 2 Yahoo! Research, Spain 3 Neo Metrics, Spain 1 Bloomberg, USA 5 Google Inc, USA 6 SAP, France 7 Avail Intelligence, Sweeden ABSTRACT At the 2010 annual ACM Conference on Recommender Systems (RecSys 2010) a panel addressed emerging topics regarding recommender systems as a whole and specifically their role in industry. This report summarizes answers from a distinguished group of industry leaders representing different industries in which recommender systems are highly relevant. Panel members discuss questions regarding the role of recommender systems in their own industry area, killer applications, opportunities, and future directions. Categories and Subject Descriptors H.3.3 [Information Storage and Retrieval]: Information Search and Retrieval – information filtering General Terms Algorithms, Management, Measurement, Performance, Design, Human Factors. Keywords Recommender Systems, Industry, Collaborative Filtering, Search 1. INTRODUCTION Recommender systems (RS) have been a fertile research area in the past 15 years [1], with collaborative filtering (CF) [4] approaches gradually becoming more popular than traditional content-based (CB) [8] recommenders, and hybrid RS [3] becoming common as well. As the volumes of information people are exposed to continue to grow dramatically, the importance of RS is likely to continue to grow and play a key role in many different industry domains. A few early examples of applications were introduced in [9]: GroupLens [7] applies CF for recommending news articles; Fab [2] is a hybrid CF-CB RS for web pages; and ReferralWeb [6] combines social networks and CF to recommend people and communities and improve search results through these recommendations. RS have evolved to become an important business tool that is reshaping the world of e-commerce [10], helping customers find and purchase products, such as songs, books, movies, or restaurants. With the emergence of the social web, RS are starting to appear in leading social media services, taking advantage of folksonomies and social networks formed by the crowd [5]. In this paper, we discuss the state-of-the-art of RS in industry, as well as directions and opportunities towards the future. Panel participants span large and small companies, serving different roles within these companies, applying both online and offline recommender technologies, and in different fields and scales. They refer to questions around the key aspects making a RS successful, the relation to search engines, the use and value of RS within their organizations and the way they are evaluated, the potential killer application, and finally enumerate the most important business and technological challenges for RS looking forward. 2. PANEL QUESTIONS 2.1 Where do you position the area of RS w.r.t existing domains - IR, HCI, data mining? What other areas play key role for RS? Henrik Schinzel. We are entering the age of recommendations. We will go from a world were text logical search is ubiquitous in finding content, to a world were recommendations will be the key technology in finding content. Thus recommendation systems will be key in information retrieval and finding content. An area I predict that RS will have a tremendous impact is advertising. Pat Moore. I have found that IR, HCI and machine learning all play a key role in helping RS to perform better, but in the area of news personalization I have found performance to be more sensitive to IR methodology. I have seen approaches with strong IR methodology and no data mining with a particular HCI treatment outperform average IR methodology with strong data mining having the same HCI treatment. Palash Nandy. I would position RS in between data mining and HCI. Chahab Nastar. Personalization and context-awareness are also central to building relevant RS. Personalization is about customizing the system w.r.t user profiles in general; context- awareness is an additional layer – customization is w.r.t data, sessions or users in particular. Note that personalization and Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. RecSys’10, September 26–30, 2010, Barcelona, Spain. Copyright 2010 ACM 978-1-60558-906-0/10/09...$10.00. 7