by Òscar Celma (BMAT) and Paul Lamere (The Echo Nest); slides
The world of music is changing rapidly.
We are now just a few clicks away from being able to listen to nearly any song that has ever been recorded.
This easy access to a nearly endless supply of music is changing how we explore, discover, share and experience music.
As the world of online music grows, music recommendation and discovery tools become an increasingly important way for music listeners to engage with music. Commercial recommenders such as Last.fm, iTunes Genius and Pandora have enjoyed commercial and critical success. But how well do these systems really work? How good are the recommendations? How far into the "long tail" do these recommenders reach?
In this tutorial we look at the current state-of-the-art in music recommendation and discovery. We examine current commercial and research systems, focusing on the advantages and the disadvantages of the various recommendation strategies. We look at some of the challenges in building music recommenders and we explore some of the novel techniques that are being used to improve future music recommendation and discovery systems.
Òscar Celma is the Chief Innovation Officer at Barcelona Music and Audio Technologies (BMAT).
In 2008, Òscar obtained his Ph.D. in Computer Science and Digital Communication, in the Pompeu Fabra University (Barcelona, Spain).
Òscar has a book published by Springer,
titled "Music Recommendation and Discovery: The Long Tail, Long Fail and Long Play in the Music Digital Age" (2010).
He holds 2 patents (US2003009344 and JP2003323188, 2002) from his work on the Vocaloid system, a singing voice-synthesizer bought by Yamaha in 2004.
Follow on Twitter: @ocelma
Paul Lamere is the Director of Developer Platform for The Echo Nest, a music intelligence company located in Boston.
Paul is interested in using technology to help people explore for new and interesting music.
He is active in both the music information retrieval and the recommender systems research communities.
Paul authors a popular blog on music technology at MusicMachinery.com.
Follow on Twitter: @plamere
by Neil Hurley (UCD); slides
The possibility of designing user rating profiles to deliberately and maliciously manipulate the recommendation output of a collaborative filtering system was first raised in 2002. One scenario proposed was that an author, motivated to increase recommendations of his book, might create a set of false profiles that rate the book highly, in an effort to artificially promote the ratings given by the system to genuine users. Since then, these attacks have been dubbed as shilling attacks or profile injection attacks. Several attack models have been proposed and the performance of these attacks in terms of influencing the system predictions has been evaluated for a number of memory-based and model-based collaborative filtering algorithms. Moreover, strategies have been proposed to enhance the robustness of existing algorithms and new algorithms have been proposed with built-in attack resistance. This tutorial will review the work that has taken place in the last decade on robustness of recommendation algorithms and seek to examine the question of the importance of robustness in future research.
Neil Hurley graduated with a M.Sc. degree in Mathematical science from University College Dublin (UCD) in 1988. In 1989, he joined Hitachi Dublin Laboratory (HDL), a computer science research laboratory based at Trinity College Dublin. During his time at HDL he worked on various topics including knowledge-based engineering, for which he was awarded a PhD in 1995, and parallel computing. He joined the academic staff in the UCD School of Computer Science and Informatics in 1999, where he has carried out research in topics including information hiding in digital content, distributed systems and graph and network analysis. He started research on recommender systems in 2002 when he first started to look at robustness issues in kNN collaborative filtering algorithms. Since then, he has worked on various topics in recommendation including recommendation diversity. To date, three students have completed their PhDs on topics in recommendation under his supervision. Currently, his major research focus is social network analysis in the Clique Network and Graph Analysis cluster.
by Daniel Tunkelang (LinkedIn) slides
Recommender systems aim to provide users with products or content that satisfy the users' stated or inferred needs. The primary evaluation measures for recommender systems emphasize either the perceived relevance of the recommendations or the actions associated with those recommendations (e.g., purchases or clicks). Unfortunately, this transactional emphasis neglects how users interact with recommendations in the context of information seeking tasks. The effectiveness of this interaction determines the user's experience beyond a single transaction. This tutorial explores the role of recommendations as part of a conversation between the user and an information seeking system. The tutorial does not require any special background in interfaces or usability, and will focus on practical techniques to make recommender systems most effective for users.
Daniel Tunkelang is a Principal Data Scientist at LinkedIn, where he leads a team addressing large-scale problems in information retrieval,
recommender systems, and network analysis. He is a leading advocate of human-computer information retrieval (HCIR):
he established the annual HCIR workshop in 2007.
He published the first textbook on faceted search, an area to which he made seminal contributions as a founder and Chief Scientist of Endeca.
Daniel also worked at Google, leading a team to improve local search quality.
Daniel hold degrees in math and computer science from MIT and a PhD in computer science from CMU, where he worked on information visualization. He blogs at The Noisy Channel.