Martin Pichl

#nowplaying Music Dataset: Extracting Listening Behavior from Twitter

The extraction of information from online social networks has become popular in both industry and academia as these data sources allow for innovative applications. However, in the area of music recommender systems and music information retrieval, respective data is hardly exploited. In this paper, we present the #nowplaying dataset, which leverages social media for the creation of a diverse and constantly updated dataset, which describes the music listening behavior of users. For the creation of the dataset, we rely on Twitter, which is frequently facilitated for posting which music the respective user is currently listening to. From such tweets, we extract track and artist information and further metadata. The dataset currently comprises 49 million listening events, 144,011 artists, 1,346,203 tracks and 4,150,615 users which makes it considerably larger than existing datasets.

Improving Context-Aware Music Recommender Systems: Beyond the Pre-filtering Approach

Over the last years, music consumption has changed fundamentally: people switch from private, mostly limited music collections to huge public music collections provided by music streaming platforms. Thus, the amount of available music has increased dramatically and music streaming platforms heavily rely on recommender systems to assist users in discovering music they like. Incorporating the context of users has been shown to improve the quality of recommendations. Previous approaches based on pre-filtering suffered from a split dataset. In this work, we present a context-aware recommender system based on factorization machines that extracts information about the users context from the names of the users playlists. Based on a dataset comprising 15,000 users and 1.8 million tracks we show that our proposed approach outperforms the pre-filtering approach substantially in terms of accuracy of the computed recommendations.

An Empirical Evaluation of Property Recommender Systems for Wikidata and Collaborative Knowledge Bases

The Wikidata platform is a crowdsourced, structured knowledgebase aiming to provide integrated, free and language-agnostic facts which are---amongst others---used by Wikipedias. Users who actively enter, review and revise data on Wikidata are assisted by a property suggesting system which provides users with properties that might also be applicable to a given item. We argue that evaluating and subsequently improving this recommendation mechanism and hence, assisting users, can directly contribute to an even more integrated, consistent and extensive knowledge base serving a huge variety of applications. However, the quality and usefulness of such recommendations has not been evaluated yet. In this work, we provide the first evaluation of different approaches aiming to provide users with property recommendations in the process of curating information on Wikidata. We compare the approach currently facilitated on Wikidata with two state-of-the-art recommendation approaches stemming from the field of RDF recommender systems and collaborative information systems. Further, we also evaluate hybrid recommender systems combining these approaches. Our evaluations show that the current recommendation algorithm works well in regards to recall and precision, reaching a recall@7 of 79.71% and a precision@7 of 27.97%. We also find that generally, incorporating contextual as well as classifying information into the computation of property recommendations can further improve its performance significantly.

#Nowplaying on #Spotify: Leveraging Spotify Information on Twitter for Artist Recommendations

The rise of the web enabled new distribution channels like online stores and streaming platforms, offering a vast amount of different products. For helping customers finding products according to their taste on those platforms, recommender systems play an important role. Besides focusing on the computation of the recommendations itself, in literature the problem of a lack of data appropriate for research is discussed. In order to overcome this problem, we present a music recommendation system exploiting a dataset containing listening histories of users, who posted what they are listening to at the moment on the microblogging platform Twitter. As this dataset is updated daily, we propose a genetic algorithm, which allows the recommender system to adopt its input parameters to the extended dataset. In the evaluation part of this work, we benchmark the presented recommender system against two baseline approaches. We show that the performance of our proposed recommender is promising and clearly outperforms the baseline.

Understanding Playlist Creation on Music Streaming Platforms

Music streaming platforms enable people to access millions of tracks using computers and mobile devices. The latter allow users consume different music during different activities. Both, the sheer amount of music and the mobile access to music makes music organization an interesting topic for multimedia researchers. Assisting users to organize their music and make the music they like easily available in the right moment, contributes to increased usability of music streaming platforms. To get a deeper understanding of how users organize music nowadays, we analyze user-created playlists crawled from the music streaming platform Spotify. Using this new data set we find an explanation of differences in the playlists using audio features and based on this compute playlist clusters. We find that 91% of all users create at least one playlist in the “feel good music”-cluster and classical music or rap music can be considered as niche music with respect to the number of playlists, however not as niche music when considering the number of users. To foster research in this field, we make our analysis tool publicly available.

Culture-Aware Music Recommendation

Integrating information about the listeners cultural background when building music recommender systems has recently been identified as a means to improve recommendation quality. In this paper, we therefore propose a novel approach to jointly model users by the users musical preferences and his/her cultural background. We describe the musical preferences of users by relying on the acoustic features of the songs the users have been listening to and characterize the cultural background of users by cultural and socio-economic features that we infer from the users country. To evaluate the impact of the proposed user model on recommendation quality, we integrate the model into a culture-aware music recommender system. We show that incorporating both acoustic information of the tracks a user has listened to as well as the cultural background of users in the form of a music-cultural user model contributes to improved recommendation performance.

HOT: A Height Optimized Trie Index for Main-Memory Database Systems

We present the Height Optimized Trie (HOT), a fast and space-efficient in-memory index structure. The core algorithmic idea of HOT is to dynamically vary the number of bits considered at each node, which enables a consistently high fanout and thereby good cache efficiency. The layout of each node is carefully engineered for compactness and fast search using SIMD instructions. Our experimental results, which use a wide variety of workloads and data sets, show that HOT outperforms other state-of-the-art index structures for string keys both in terms of search performance and memory footprint, while being competitive for integer keys. We believe that these properties make HOT highly useful as a general-purpose index structure for main-memory databases.

Understanding User-Curated Playlists on Spotify: A Machine Learning Approach

Musicstreamingplatformsenablepeopletoaccessmillionsoftracksusingcomputersandmobile devices.However,userscannotbrowsemanuallymillionsoftrackstofindmusictheylike.Building recommendersystemssuggestingmusicfittingthecurrentcontextofauserisachallengingtask. A deeper understanding for the characteristics of user-curated playlists naturally contributes to morepersonalizedrecommendations.Togetadeeperunderstandingofhowusersorganizemusic nowadays,weanalyzeuser-curatedplaylistsfromthemusicstreamingplatformSpotify.Basedon theaudiofeaturesofthetracks,wefindanexplanationofdifferencesintheplaylistsusingaPCA andareabletogroupplaylistsusingspectralclustering.Ourfindingsaboutplaylistcharacteristics canbeexploitedinaSVD-basedmusicrecommendersystemandourproposedclusteringapproach forfindinggroupsofsimilarplaylistsiseasytointegrateintoarecommendersystemusingpre-or post-filteringtechniques. KEywoRDS Clustering, Data Acquisition, Data Analysis, Information Retrieval, Machine Learning, Music Information Retrieval, Quantitative Study, User Modeling, User-Generated Content