How the Semantic Web Will Redefine Digital Music
Monday’s SanFran MusicTech Summit featured a host of informative panels dealing with the ways technology is transforming the music industry. One thread concerned how music companies are using the so-called “Semantic Web,” the field of web data from which actual meaning can be derived through analysis and machine learning. Most prominently, recommendation engines, messaging startups, marketing organizations and data reporting agencies are all using the Semantic Web to improve their understanding of both music content and artist reputations, improving the knowledge base of an industry known for traditionally poor market research and analysis.
Two startups that presented during an afternoon demo session have deep roots in the Semantic Web. Boston’s The Echo Nest touted its “music brain,” a recommendation platform that draws on both sonic analysis and text mentions around the web, including sentiment analysis, while Atlanta-based Band Metrics said it draws on social networking content, blog mentions, YouTube play counts and other information to gauge the popularity of artists and report on their reputations in real time. (Both also made announcements: The former will be used to power European streaming music site Spotify, while the latter introduced new formatting standards for social networks to report data.)
Elsewhere at the event, discussions hinted at the importance of the Semantic Web, if they didn’t always acknowledge it overtly. According to a morning panel, sentiment analysis — the form of natural language processing that assigns positive or negative attributes to a statement — can prove useful to a record label, music marketer or consumer brand looking to gauge listeners’ interest or feelings about an artist. A late-afternoon panel suggested that traditional recommendation engines, which have typically been built by editorial experts or used collaborative filtering, can be improved by harvesting relevant data from a massive river of unstructured information, helping to prevent the engines from being gamed by hackers or misled by feedback loops.
Innovators in the field are positioned to capitalize on the continuing trends toward a more even balance between the web’s read and write functions, more disaggregated content and more self-published social objects that must be searched to be understood fully. The information they gather can potentially be used to empower worthwhile lesser-known artists by helping them reach new audiences, deliver better targeted advertising, and save record labels and consumer brands a lot of money and trouble in the event that an icon’s reputation suddenly turns sour. Analysis of trending subject matter can even be used as a predictive engine. And while a game-changing company that turns the music industry on its ear may have yet to emerge, the appearance of startups using the Semantic Web points the way toward a smarter industry in the future, one that learns how to read the tea leaves hidden in web data for its benefit.
Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.
SF MusicTech was a fantastic conference and it’s great to see you idenitfy one of the most important threads. Tools like BandMetrics promise to become invaluable to managers and their artists. Knowing what to do with it, of course, will be critical to taking advantage of this new data stream.
Wow — sounds like an amazing event. So sorry to have missed it; a number of the same themes are under discussion here at the Web 3.0 Conference in New York.
Intrigued and excited to learn more about the Echo Nest and Band Metrics.
It better beat what’s capable with today’s folksonomy. Every bit of chamber music tagged with “Japanese” and “toast” makes the system useless. Users are too stupid to be effective librarians.
for music content recommendations via unstructured analysis – are only as good as the metadata models. blind content classification methods circa 99 coupled with semtech and predictive analytics circa 2007-2009 will be the answer. apple knows the importance of metadata accuracy. it is good to see this stuff coming full circle.
@jaxsoncreole