The paper I. Vatolkin – Exploration of Two-Objective Scenarios on Supervised Evolutionary Feature Selection: a Survey and a Case Study (Application to Music Categorisation) is accepted for EMO 2015. The conference will take place on 29.3-1.4.2015 in Guimarães, Portugal.
The article G. Bonnin and D. Jannach – Automated Generation of Music Playlists: Survey and Experiments, ACM Computing Surveys, Vol. 47, No. 2, Article 26 is now available online.
The paper is a detailed survey on playlist generation. First, different generation algorithms and the different types of data they can use are classified, the common approaches for the evaluation of the quality of the generated playlists are reviewed. In a second part, the results of thorough experiments are presented in which several existing algorithms are compared using several evaluation approaches. The results suggest that the order of the tracks, artist information, and track popularity are three important criteria to take into account when generating playlists and that some existing evaluation approaches have their limitations.
As a part of Digital Entertainment Technologies student project, various mobile music games will be developed for Android platform using Unity in small groups of 3-5 students. Different features from actual MIR research (audio signal characteristics, high-level semantic descriptors, playlist statistics) should be integrated to control the game process. The results will be presented on 15.12.2014.
The paper on the analysis of playlist characteristics (D. Jannach, I. Kamehkhosh, and G. Bonnin – Analyzing the Characteristics of Shared Playlists for Music Recommendation) is accepted for the 6th ACM RecSys Workshop on Recommender Systems and Social Web.
At the opening session of European Conference on Data Analysis 2014 in Bremen, Daniel Stoller received one of two “Best Paper” awards for ECDA 2013 postproceeding paper, “Impact of Frame Size and Instrumentation on Chroma-based Automatic Chord Recognition”.
With Daniel as the first author, this paper presents the results of the joint work together with Matthias Mauch (Queen Mary University of London), Igor Vatolkin (TU Dortmund, Department of Computer Science), and Claus Weihs (TU Dortmund, Faculty of Statistics). Daniel works as a student assistant at the Chair of Algorithm Engineering, Department of Computer Science, TU Dortmund. For further details about the awards see the book of abstracts.
12th Workshop on Intelligent Techniques for Web Personalization & Recommendation (ITWP 2014)
In conjunction with AAAI 2014, July 27/28, 2014 – Quebec, Canada
Dietmar Jannach, Department of Computer Science, Technische Universität Dortmund, Germany
Bamshad Mobasher, School of Computer Science, DePaul University, Chicago, USA
* 10 April: Paper submission deadline
* 1 May: Author notification
* 15 May: Camera-ready versions due
* 27/28 July: Workshop held
For further details please visit http://ls13-www.cs.uni-dortmund.de/homepage/itwp2014/
The music database for genre and style recognition has been extended with an artist-independent test set TSAI120. This set has the same genre and approximately the same style distribution and is designed to evaluate the possible overfitting towards particular artists / albums. The description of the complete database and the lists of features are here.
The songs can not be freely distributed. However, if you are interested in audio features from my PhD, you can write me an email!
The PhD is available online:
Bernd Bischl – Model and Algorithm Selection in Statistical Learning and Optimization.
I would like to announce a Session “Statistical Musicology” at the 38th annual conference of the German Classification Society (GfKl), called the 2nd European Conference on Data Analysis, July 2 – 4, 2014, in Bremen (Germany).
The conferences of the GfKl are well-known for both, their theoretical and their practical contributions, on various fields of classification and data analysis. Since last year, the GfKl conferences are called European Conferences on Data Analysis. Statistical Musicology is the eleventh time on the schedule of the GfKl conference. On the first ten occasions the musicology sessions comprised between 6 and 10 papers each. Moreover, there were invited talks in the field in 2004 by Jan Beran, in 2009 by Taylan Cemgil, and in 2010 by myself at the conference.
In 2014 the session is again open to all topics in the field. The idea is to give all statistically oriented groups in the field a forum to improve communication as well as to give us the chance to also join statistical methodology oriented sessions on supervised and unsupervised learning and data analysis in general at the conference.
I would like to invite you and your group cordially to present a paper at the musicology session.
Please note that abstracts have to be submitted until
March 15, 2014 extended to March 31, 2014!!
More information and templates can be found under:
Papers corresponding to presentations can be published in post-conference proceedings. Full papers can be submitted until September 30, 2014.
I am looking forward to welcome you at the conference.
Papers on user acceptance of recommender systems may be prepared for the special issue of the Journal of Personalization Research. The deadline for extended abstract submission is February 1, 2014.
For further details see http://ls13-www.cs.uni-dortmund.de/homepage/umuai2013/index.shtml.