Streamwatchr — Inside the world's playlist

Manos Tsagkias, Wouter Weerkamp, and Maarten de Rijke

University of Amsterdam
01 January 2016
Keywords: social media, music, named entity linking, lyrics extraction, real-time trends

Abstract

Streamwatchr (2013–2016), developed at the University of Amsterdam, leveraged machine learning and named entity extraction to track global music listening behavior in real time through tweets. It offered features like Top-100 charts, interactive maps, and a “radio” stream mode driven by a dynamic recommender system. Using Python and MongoDB, it processed over 438 million tweets, identified 660,941 artists, and linked them to MusicBrainz and YouTube videos. This innovative blend of analytics and entity linking earned it a spot in the Dutch delegation at South by Southwest (SXSW).

Streamwatchr (2013–2016) was a web application developed at the University of Amsterdam, by Wouter Weerkamp, Maarten de Rijke, and myself. Streamwatchr monitored the Twitter feed for #nowplaying, and tracked how people interacted with songs, artists, albums, in real-time. In its latest version it also identified which parts of the lyrics were the most sung along!

Streamwatchr’s innovative approach earned it a place in the Dutch delegation for innovation at the internationally renowned South by Southwest (SXSW) festival.

Screencast of Streamtchr on YouTube. The video covers all the features of the platform.

Streamwatchr provided unique insights into global music listening behavior through features such as Top-100 charts, interactive maps, and a “radio” stream mode. The radio stream played a sequence of YouTube videos, with each song following the next based on the probability of them being played together. This was achieved by linking identified songs to YouTube videos and generating playlists driven by a recommender system. Behind the scenes, the system utilized a directed graph, where nodes represented songs, and edge weights encoded the likelihood of transitioning from one song to another. These weights were dynamically updated with every incoming tweet.

The engineering behind Streamwatchr was equally groundbreaking [1]. It collected music-related tweets in real time, extracted song and artist information, and mapped them to MusicBrainz, a comprehensive music database, and corresponding YouTube video clips. Every aspect of Streamwatchr, from popularity charts and trending music to song recommendations and analytics, was refreshed with each new tweet, ensuring a dynamic and up-to-date user experience.

Streamwatchr leveraged a tech stack of Python and MongoDB for analytics and recommendations. Its real-time capabilities were powered by efficient data structures and algorithms that minimized the computational overhead required for updates, delivering a seamless and responsive experience.

The millions of tweets and the hundreds of thousands of artists that Streamwatchr has listened to over the years have been distilled to a handful of noteworthy factoids:

Factoid
tweets listened 438,225,941
artists seen 660,941
most popular song Passenger – Let Her Go, 196,986 times
most sung along song John Legend – All of Me, 541 times

References

[1]: Wouter Weerkamp, Manos Tsagkias, and Maarten de Rijke. Inside the world’s playlist. In International Conference of Knowledge Management (ICKM) 2013. ACM Library; PDF

Related Posts

  1. Abstract

    In this paper we describe how Twitter is used in various languages. We observe notable differences between languages regarding the use of hashtags, links, mentions, and conversations. We propose two dimensions that can be used to classify languages, each of which is likely to require different ways of analysis.

    References

    [1] Wouter Weerkamp, Simon Carter, and Manos Tsagkias. 2011. How People use Twitter in Different Languages. In Web Science 2011. UVA Link; PDF.

    23.95% similar — Paper at WebScience 2011 — How People use Twitter in Different Languages

  2. References

    [1] Manos Tsagkias, Maarten de Rijke, and Wouter Weerkamp. 2011. Linking online news and social media. In Proceedings of the fourth ACM international conference on Web search and data mining (WSDM ‘11). Association for Computing Machinery, New York, NY, USA, 565–574. ACM Link PDF

    23.67% similar — Paper at WSDM 2011 — Linking online news and social media

  3. The popularity of microblogging platforms, such as Twitter, renders them valuable real-time information resources for tracking various aspects of worldwide events, e.g., earthquakes, political elections, etc. Such events are usually characterized in microblog posts via the use of hashtags (#). As microbloggers come from different backgrounds, and express themselves in different languages, we witness different “translations” of hashtags which, however, are about the same event. Language-dependent variants of hashtags can possibly lead to issues in content-analysis. In this paper, we propose a method for translating hashtags, which builds on methods from information retrieval. The method introduced is source and target language independent. Our method is desirable, either instead of, or complimentary, to the direct translation of the hashtag for three reasons. First we return a list of hashtags on the same topic, which takes into account the plurality and variability of hashtags used by

    23.65% similar — Paper at WebScience 2011 — Twitter hashtags: Joint Translation and Clustering

  4. Abstract

    Offering access to information in microblog posts requires successful language identification. Language identification on sparse and noisy data can be challenging. In this paper we explore the performance of a state-of-the-art n-gram-based language identifier, and we introduce two semi-supervised priors to enhance performance at microblog post level: (i) blogger-basedprior, using previous posts by the same blogger, and (ii) link-based prior, using the pages linked to from the post. We test our models on five languages (Dutch, English, French, German, and Spanish), and a set of 1,000 tweets per language. Results show that our priors improve accuracy, but that there is still room for improvement.

    References

    23.26% similar — Paper at DIR 2011 — Semi-Supervised Priors for Microblog Language Identification

  5. ```

    22.61% similar — Ph.D. Thesis — Mining Social Media: Tracking Content and Predicting Behavior

  6. We describe the participation of the University of Amsterdam’s ILPS group in the web, blog, web, entity, and relevance feedback track at TREC 2009. Our main preliminary conclusions are as follows. For the Blog track we find that for top stories identification a blogs to news approach outperforms a simple news to blogs approach. This is interesting, as this approach starts with no input except for a date, whereas the news to blogs approach also has news headlines as input. For the web track, we find that spam is an important issue in the ad hoc task and that Wikipedia- based heuristic optimization approaches help to boost the retrieval performance, which is assumed to potentially reduce the spam in top ranked documents. As for the diversity task, we explored different methods. Initial results show that clustering and a topic model-based approach have similar performance, which are relatively better than a query log based approach. Our performance in the Entity track was downright

    20.58% similar — Paper at TREC 2009 — The University of Amsterdam at TREC 2009

  7. Abstract

    We predict IMDb movie ratings and consider two sets of features: surface and textual features. For the latter, we assume that no social media signal is isolated and use data from multiple channels that are linked to a particular movie, such as tweets from Twitter and comments from YouTube. We extract textual features from each channel to use in our prediction model and we explore whether data from either of these channels can help to extract a better set of textual feature for prediction. Our best performing model is able to rate movies very close to the observed values.

    19.99% similar — Paper at ECIR 2012 — Predicting IMDB Move Ratings Using Social Media