Exploiting Social Media for Music Information Retrieval
This chapter will first provide an introduction to information retrieval (IR) in general, before briefly explaining the research field of music information retrieval (MIR). Hereafter, we will discuss why and how social media mining (SMM) techniques can be
- PDF / 593,636 Bytes
- 30 Pages / 439.36 x 666.15 pts Page_size
- 35 Downloads / 216 Views
Abstract This chapter will first provide an introduction to information retrieval (IR) in general, before briefly explaining the research field of music information retrieval (MIR). Hereafter, we will discuss why and how social media mining (SMM) techniques can be beneficially employed in the context of MIR. More precisely, motivations for the common MIR tasks of music similarity computation, music popularity estimation, and auto-tagging music will be provided, and the current state-of-the-art in employing SMM techniques to these three tasks will be elaborated. Developing music similarity measures is an important task in MIR as such measures are a key ingredient for music recommendation systems, automated playlist generators, and intelligent browsing interfaces, among others. In this chapter, it will be shown how to infer music similarity information from microblogs, collaborative tags, web pages, playlists, and peer-to-peer networks. Estimating the popularity of a music item is obviously important for the music industry but also to create serendipitous music retrieval and recommendation systems. Therefore, approaches that derive such information from web page counts, geo-located microblogs, a peerto-peer network, and a social music platform will be reviewed. Eventually, different music auto-tagging methods that assign semantic labels to music pieces will be presented. In particular, computational approaches that rely on machine learning techniques as well as human-centred strategies that infer tags directly from some kind of user input (e.g. “games with a purpose”) will be addressed.
M. Schedl () Department of Computational Perception, Johannes Kepler University, Altenberger Straße 69, 4040 Linz, Austria e-mail: [email protected] N. Ramzan et al. (eds.), Social Media Retrieval, Computer Communications and Networks, DOI 10.1007/978-1-4471-4555-4 20, © Springer-Verlag London 2013
449
450
M. Schedl
1 Introduction to Information Retrieval The discipline of information retrieval (IR) is a mature field of research as early work dates back to the 1950s, for instance [59]. Since I can only give a very brief introduction to this exciting field here, the interested reader is referred to one of the many excellent books that offer comprehensive coverage of IR. I personally recommend [22] for an introduction and [3] and [8] for a more comprehensive coverage. Broadly speaking, IR is concerned with elaborating and testing methods to uncover information from potentially large corpora of text (traditional IR) or (more recently) multimedia, in response to the user’s expression of an information need. This information need is usually given as a text query, the classical example being a user who types in a query string into his or her preferred search engine. Texts are most frequently organised in the form of documents, although other representations exist. Hence, it is usually also documents which are returned as response to a query to a search engine. In order to be able to promptly provide search results for millions of
Data Loading...
