{"id":198,"date":"2011-12-01T14:11:24","date_gmt":"2011-12-01T13:11:24","guid":{"rendered":"http:\/\/www.schnitzer.at\/dominik\/?page_id=198"},"modified":"2024-01-02T20:21:34","modified_gmt":"2024-01-02T19:21:34","slug":"phd-thesis","status":"publish","type":"page","link":"https:\/\/www.schnitzer.at\/dominik\/phd-thesis\/","title":{"rendered":"PhD Thesis: Indexing Content-Based Music Similarity Models for Fast Retrieval in Massive Databases"},"content":{"rendered":"

Indexing Content-Based Music Similarity Models for Fast Retrieval in Massive Databases<\/h3>\n

~ Dealing with the Music of the World ~ <\/em>
\nDownload PhD Thesis<\/a>
\nDefense Presentation (Jan. 31, 2012)<\/a><\/p>\n

This thesis develops a large-scale music recommendation system. Three problems are solved preventing the currently top-performing class of content-based music similarity algorithms from being used as recommendation engine in huge databases with millions of songs.<\/p>\n

    \n
  1. It is shown how to correctly use their non-vectorial music similarity features with their non-metric divergences in centroid-computing algorithms.<\/em><\/li>\n
  2. An alleviation to the problem of “hubs” is presented.<\/em><\/li>\n
  3. A method to speed up music recommendation queries is developed.<\/em><\/li>\n<\/ol>\n

    All three methods are merged in a large-scale, high-quality music recommendation prototype. The prototype is called “Wolperdinger” and operates on a collection of 2.3 million songs. A query is processed in a fraction of a second on a standard PC.<\/p>\n

    Errata<\/h4>\n

    ~~~<\/span><\/p>\n

    Video of the Prototype<\/h4>\n

    The video shows a music recommendation system operating on 2.3 million songs.
    \n