Showing papers by "J. Stephen Downie published in 2014"

A Musical Progression with Greenstone: How Music Content Analysis and Linked Data is Helping Redefine the Boundaries to a Music Digital Library

Abstract: Despite the recasting of the web's technical capabilities through Web 2.0, conventional digital library software architectures---from which many of our leading Music Digital Libraries (MDLs) are formed---result in digital resources that are, surprisingly, disconnected from other online sources of information, and embody a "read-only" mindset. Leveraging from Music Information Retrieval (MIR) techniques and Linked Open Data (LOD), in this paper we demonstrate a new form of music digital library that encompasses management, discovery, delivery, and analysis of the musical content it contains. Utilizing open source tools such as Greenstone, audioDB, Meandre, and Apache Jena we present a series of transformations to a musical digital library sourced from audio files that steadily increases the level of support provided to the user for musicological study. While the seed for this work was motivated by better supporting musicologists in a digital library, the developed software architecture alters the boundaries to what is conventionally thought of as a digital library---and in doing so challenges core assumptions made in mainstream digital library software design.

What is this song about anyway?: automatic classification of subject using user interpretations and lyrics

Abstract: Metadata research for music digital libraries has traditionally focused on genre. Despite its potential for improving the ability of users to better search and browse music collections, music subject metadata is an unexplored area. The objective of this study is to expand the scope of music metadata research, in particular, by exploring music subject classification based on user interpretations of music. Furthermore, we compare this previously unexplored form of user data to lyrics at subject prediction tasks. In our experiment, we use datasets consisting of 900 songs annotated with user interpretations. To determine the significance of performance differences between the two sources, we applied Friedman's ANOVA test on the classification accuracies. The results show that user-generated interpretations are significantly more useful than lyrics as classification features (p < 0.05). The findings support the possibility of exploiting various existing sources for subject metadata enrichment in music digital libraries.

Ten Years of MIREX (Music Information Retrieval Evaluation eXchange): Reflections, Challenges and Opportunities.

Abstract: (t M m r h w T e f c c i h t M m h M (t E I a (r w t s t r t a p t a l T (M to M m ra ha w T ea fl ca ch in ho ti M m ha M (M th E Il a (M re w th si th re ta an po ti ab lu S A p h O tr Th M ob MI mu an as wit Th ar le al ha ng ol o MI mu av Mu M he Ev lli f M es wh he ig he es ak nd oi o bi ut Sa Att ph ham Op rie he MI be IR us ng s th his rli ct l c al g m li n IR un ve us MI e va in fr MI su ha e t gn e r se A ke d in n ili tio lly tri en m pp ev j c R er RE sic ge in h s ie ts ch lle m st nar RE ni e si R I alu no am R ult at te nif re ea As e c nt n f it on y ib n m, por va j c M RE r 2 EX c e nv a p es s u ha en mo ti ry EX ty m c RE In ua ois m R) ts T ex fi es arc s th ch t w fr ty ns Jo bu D Y rtu al C d k Mu EX 20 X in o vo a pa st up al ng od c y, X y ma EX nte at s me r s Te xt c se ch th hi ha w ram y o s, o tio Do Yu un C J U do U ka us X 0 X n of ol m ap p p lle ge de u , s X. m an I X er tio a ew re o ex i an ea h hi is all wh m of a C on w un nit on J. U ow U ah si X) 14 h fo t lv me pe pl o en es el us so W me ny In X) rn on at wo es f xt …

Big data at scale for digital humanities: An architecture for the HathiTrust research center

Abstract: Big Data in the humanities is a new phenomenon that is expected to revolutionize the process of humanities research. The HathiTrust Research Center (HTRC) is a cyberinfrastructure to support humanities research on big humanities data. The HathiTrust Research Center has been designed to make the technology serve the researcher to make the content easy to find, to make the research tools efficient and effective, to allow researchers to customize their environment, to allow researchers to combine their own data with that of the HTRC, and to allow researchers to contribute tools. The architecture has multiple layers of abstraction providing a secure, scalable, extendable, and generalizable interface for both human and computational users. Stacy T. Kowalczyk Dominican University, USA Yiming Sun Indiana University, USA Zong Peng Indiana University, USA Beth Plale Indiana University, USA Aaron Todd Indiana University, USA Loretta Auvil University of Illinois, USA Craig Willis University of Illinois, USA Jiaan Zeng Indiana University, USA Milinda Pathirage Indiana University, USA Samitha Liyanage Indiana University, USA Guangchen Ruan Indiana University, USA J. Stephen Downie University of Illinois, USA DOI: 10.4018/978-1-4666-4699-5.ch011

A Cross-Cultural Study on the Mood of K-POP Songs.

Abstract: Prior research suggests that music mood is one of the most important criteria when people look for music—but the perception of mood may be subjective and can be influenced by many factors including the listeners’ cultural background. In recent years, the number of studies of music mood perceptions by various cultural groups and of automated mood classification of music from different cultures has been increasing. However, there has yet to be a well-established testbed for evaluating cross-cultural tasks in Music Information Retrieval (MIR). Moreover, most existing datasets in MIR consist mainly of Western music and the cultural backgrounds of the annotators were mostly not taken into consideration or were limited to one cultural group. In this study, we built a collection of 1,892 K-pop (Korean Pop) songs with mood annotations collected from both Korean and American listeners, based on three different mood models. We analyze the differences and similarities between the mood judgments of the two listener groups, and propose potential MIR tasks that can be evaluated on this dataset.

The HathiTrust Corpus: A Digital Library for Musicology Research?

Abstract: The HathiTrust Digital Library (HTDL) consists of digitized print materials contributed from the collections of some of the foremost research libraries of the world. The HTDL contains over 11 million volumes comprising approximately 3.9 billion pages. In this paper, we describe an exploratory bibliometric study to examine and characterize music-related content in the HTDL. Our study provides an overview of the music-related content in the HTDL as seen through the lenses of format, genre, language, and chronology. We seek to determine in what ways, if any, the materials in HTDL could be considered to form a unique music digital library for use by musicology scholars and students. We also suggest ways in which the music-related content of the HTDL holdings could be made more useful to users with musicological needs and interests.

Using Collections and Worksets in Large-Scale Corpora: Preliminary Findings from the Workset Creation for Scholarly Analysis Project

Abstract: Scholars from numerous disciplines rely on collections of texts to support research activities. On this diverse and interdisciplinary frontier of digital scholarship, libraries and information institutions must 1) prepare to support research using large collections of digitized texts, and 2) understand the different methods of analysis being applied to the collections of digitized text across disciplines. The HathiTrust Research Center’s Workset Creation for Scholarly Analysis (WCSA) project conducted a series of focus groups and interviews to analyze and understand the scholarly practices of researchers that use largescale, digital text corpora. This poster presents preliminary findings from that study, which offers early insights into user requirements for scholarly research with textual corpora.