DOCSLIB.ORG

Computing the Semantic Relatedness of Music Genres Using Semantic Web Data Dennis Diefenbach, Pierre-René Lherisson, Fabrice Muhlenbach, Pierre Maret

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Computing the Semantic Relatedness of Music Genres Using Semantic Web Data Dennis Diefenbach, Pierre-René Lherisson, Fabrice Muhlenbach, Pierre Maret Computing the Semantic Relatedness of Music Genres using Semantic Web Data Dennis Diefenbach, Pierre-René Lherisson, Fabrice Muhlenbach, Pierre Maret To cite this version: Dennis Diefenbach, Pierre-René Lherisson, Fabrice Muhlenbach, Pierre Maret. Computing the Se- mantic Relatedness of Music Genres using Semantic Web Data. Semantics 2016, Sep 2016, Leipzig, Germany. hal-01637065 HAL Id: hal-01637065 https://hal.archives-ouvertes.fr/hal-01637065 Submitted on 17 Nov 2017 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Computing the Semantic Relatedness of Music Genres using Semantic Web Data Dennis Diefenbach Pierre-René Lhérisson Fabrice Muhlenbach Université de Lyon, CNRS Université de Lyon, CNRS Université de Lyon, CNRS UMR 5516 Laboratoire Hubert UMR 5516 Laboratoire Hubert UMR 5516 Laboratoire Hubert Curien Curien Curien Saint-Etienne, France Saint-Etienne, France Saint-Etienne, France dennis.diefenbach@univ- pr.lherisson@univ-st- fabrice.muhlenbach@univ- st-etienne.fr etienne.fr st-etienne.fr Pierre Maret Université de Lyon, CNRS UMR 5516 Laboratoire Hubert Curien Saint-Etienne, France pierre.maret@univ-st- etienne.fr ABSTRACT To make reccomandations we therefore have to relay on the Computing the semantic relatedness between two entities few information we have about their tracks, like their music has many applications domains. In this paper, we show a genre. Moreover we face the so-called cold-start problem, new way to compute the semantic relatedness between two i.e. we cannot use user profiles to generate the recommen- resources using semantic web data. Moreover, we show how dations. this measure can be used to compute the semantic relat- While it is nearly impossible to find information about edness between music genres which can be used for music independent artists in the Semantic Web it is easy to find recommendation systems. information about the music genre. These include songs of We first describe how to build a vector representations for a particular genre, bands that play the genre, characteristic resources in an ontology. Subsequently we show how these instruments for the genre, influences, regional distributions vector representations can be used to compute the semantic and many more. relatedness of two resources. Finally, as an application, we The problem we address here is how to exploit semantic show that our measure can be used to compute the semantic web data to answer question like "How similar is Ska and relatedness of music genres. Reggae?" or "How similar is Ska and Classical Music?". The answer to these questions can then be used to recommend to a person that is listening Ska a song of a different but CCS Concepts close music genre. •Information systems ! Similarity measures; Lan- Answering the above questions is equivalent to find out guage models; what is the semantic relatedness between two music gen- res. We propose first a way to construct for resource of an Keywords ontology a vector representation of it. Using these vector representations we propose a new way to compute the se- Vector Representations, Semantic Relatedness, Music Gen- mantic relatedness of two resources in an ontology. Finally, res, Recommendation Systems we show the results we get if we apply it to music genres. The publication is organized as follows. In section 2 we 1. INTRODUCTION discuss the related work. In section 3 we describe how to The work presented in this paper is the result of a re- compute the semantic relatedness between two resources in search work on music recommendation systems conducted an ontology. In section 4 we show our results when applied in collaboration with a french startup company called 1D to music genres and finally in section 5 we discuss future Lab1. The aim is to create a music recommendation system work. over a database of independent artists, i.e. artists who are not under contract with the major labels. The independent, emerging artists, suffer from a lack of visibility on the web. 2. RELATED WORK Recommender Systems are designed to guide users through 1 http://en.1d-lab.eu large volumes of data. One of the most popular meth- ods is collaborative-filtering where the recommendation is c 2016 Copyright held by the author/owner(s). based on users' preferences and behaviors [10]. Another ap- SEMANTICS 2016: Posters and Demos Track September 13-14, 2016, Leipzig, Germany proach is content-based recommendation which uses infor- mation associated to items like unstructured text for web pages, blogs, keywords, attributes and properties for mu- resources of the ontology. The i-th component is defined as: sic, and tries to match them to the user's profile [6]. For 8 2 0 if i2 = Rr;1 [ ::: [ Rr;k example Pandora , a radio streaming platform, uses as fea- > tures tags set by experts to recommend music. Since hu- <> 1 if i 2 Rr;1 (v ) = man tagging is a time-consuming task there are methods r i . > . > to retrieve information automatically. [11] uses signal pro- : 1 if i 2 R cessing on music and inferred high-level semantic descriptors k r;k (timber, rhythm, tempo). Other meta-data provided with a i.e. it contains zero-entries for all resources, except a 1 for track like the music genre can be used. Linked Open Data the resources in Rr;1, a 1=2 for the resource Rr;2 and so on. and Encyclopedic Knowledge Source can also help learn- To compute the semantic relatedness between two resources ing more accurate items features. On the artist's level, [8] r and s we compute the cosine similarity between the two propose dbrec, a recommender system based on DBpedia. corresponding vectors vr and vs: DBpedia is also used in the movie recommendation field [7]. P They use the genres, but also information like the actors, di- i(vr)i · (vs)i < vr; vs >= rectors, writers, to create a vector for movies and compute jjvrjj · jjvsjj similarities with a Semantic version of the classical Vector This will return a number between 0 and 1. There are two Space Model (sVSM) that they created. These work require possible interpretations for this similarity measure. The first detailed information on each artist or movie that can be is that the similarity of the resources is based on the angle recommended. Since we are working with a dataset of inde- between the corresponding vector representations. The sec- pendent artists we don't find machine readable data for most ond is to interpret it as the overlap of some sub-graphs of of them in the Linked Data Cloud. We therefore choose to G. The vector of r contains non-zero entries for all resources use the music genre that is provided with our data to make that are at a distance of maximum k from r. Let G be the recommendations. We have seen in [2] that there is no big r sub-graph containing these resources. Analogously for the difference between a content-based semantic distance and a vector of s. This means that the sum in the cosine similar- simple genre-based baseline. ity is non-zero for all resources that are at a distance of k We therefore concentrated on the problem of using seman- from both r and s, i.e. we compute how big is the overlap tic data to find the semantic relatedness of music genres. between the sub-graphs G and G . There are different works that propose semantic relatedness r s measures over ontologies [4, 9]. The first proposes a se- mantic relatedness based on the top-k path connecting two 4. SEMANTIC RELATEDNESS OF MUSIC resources. First the top-k paths are found. Then the re- GENRES latedness is computed based on the properties appearing in these paths. The second is based on the number of differ- As an application of our similarity measure we have com- puted the similarity between music genres. As an ontology ent types of connection that exits between two resources x 3 and y. It considers direct connections between x and y and we considered the English DBpedia ontology. We created a connections where a third resource z is connected to x and list of music genre. Many of them could be identified using y with the same relation. the dbo:MusicGenre class. In a next step we computed the Here we propose a new semantic relatedness measure. As vector representation of all music genre. We constructed the we have shown in [3] our method has the advantage of con- vectors as described above by setting k = 2 and by consid- sidering all relations and classes of the ontology. Also, notice ering the ontology graph as an undirected graph. Note that that [9] considers distance between artist in DBpedia, while for a music genre r the corresponding vector-representation we are considering music genres, because as previously said, contains non-zero entries for example for: instruments that lots of independent artists are not described on the Web nor are characteristic for the genre, the region where it comes in the Semantic Web. from and it's stylistic origin. Moreover since we consider the graph as undirected it contains also bands that play this genre, albums and tracks that are of this genre and many more.
Load more

Recommended publications
  • 26 in the Mid-1980'S, Noise Music Seemed to Be Everywhere Crossing
    In the mid-1980’s, Noise music seemed to be everywhere crossing oceans and circulating in continents from Europe to North America to Asia (especially Japan) and Australia. Musicians of diverse background were generating their own variants of Noise performance. Groups such as Einstürzende Neubauten, SPK, and Throbbing Gristle drew larger and larger audiences to their live shows in old factories, and Psychic TV’s industrial messages were shared by fifteen thousand or so youths who joined their global ‘television network.’ Some twenty years later, the bombed-out factories of Providence, Rhode Island, the shift of New York’s ‘downtown scene’ to Brooklyn, appalling inequalities of the Detroit area, and growing social cleavages in Osaka and Tokyo, brought Noise back to the center of attention. Just the past week – it is early May, 2007 – the author of this essay saw four Noise shows in quick succession – the Locust on a Monday, Pittsburgh’s Macronympha and Fuck Telecorps (a re-formed version of Edgar Buchholtz’s Telecorps of 1992-93) on a Wednesday night; one day later, Providence pallbearers of Noise punk White Mice and Lightning Bolt who shared the same ticket, and then White Mice again. The idea that there is a coherent genre of music called ‘Noise’ was fashioned in the early 1990’s. My sense is that it became standard parlance because it is a vague enough category to encompass the often very different sonic strategies followed by a large body of musicians across the globe. I would argue that certain ways of compos- ing, performing, recording, disseminating, and consuming sound can be considered to be forms of Noise music.
    [Show full text]
  • Music on PBS: a History of Music Programming at a Community Radio Station
    Music on PBS: A History of Music Programming at a Community Radio Station Rochelle Lade (BArts Monash, MArts RMIT) A thesis submitted for the degree of Doctor of Philosophy January 2021 Abstract This historical case study explores the programs broadcast by Melbourne community radio station PBS from 1979 to 2019 and the way programming decisions were made. PBS has always been an unplaylisted, specialist music station. Decisions about what music is played are made by individual program announcers according to their own tastes, not through algorithms or by applying audience research, music sales rankings or other formal quantitative methods. These decisions are also shaped by the station’s status as a licenced community radio broadcaster. This licence category requires community access and participation in the station’s operations. Data was gathered from archives, in‐depth interviews and a quantitative analysis of programs broadcast over the four decades since PBS was founded in 1976. Based on a Bourdieusian approach to the field, a range of cultural intermediaries are identified. These are people who made and influenced programming decisions, including announcers, program managers, station managers, Board members and the programming committee. Being progressive requires change. This research has found an inherent tension between the station’s values of cooperative decision‐making and the broadcasting of progressive music. Knowledge in the fields of community radio and music is advanced by exploring how cultural intermediaries at PBS made decisions to realise eth station’s goals of community access and participation. ii Acknowledgements To my supervisors, Jock Given and Ellie Rennie, and in the early phase of this research Aneta Podkalicka, I am extremely grateful to have been given your knowledge, wisdom and support.
    [Show full text]
  • Networks of Music Groups As Success Predictors
    Networks of Music Groups as Success Predictors Dmitry Zinoviev Mathematics and Computer Science Department Suffolk University Boston, Massachussets 02114 Email: dzinoviev@suffolk.edu Abstract—More than 4,600 non-academic music groups process of penetration of the Western popular music culture emerged in the USSR and post-Soviet independent nations in into the post-Soviet realm but fails to outline the aboriginal 1960–2015, performing in 275 genres. Some of the groups became music landscape. On the other hand, Bright [12] presents an legends and survived for decades, while others vanished and are known now only to select music history scholars. We built a excellent narrative overview of the non-academic music in the network of the groups based on sharing at least one performer. USSR from the early XXth century to the dusk of the Soviet We discovered that major network measures serve as reason- Union—but with no quantitative analysis. Sparse and narrowly ably accurate predictors of the groups’ success. The proposed scoped studies of few performers [13] or aspects [14] only network-based success exploration and prediction methods are make the barren landscape look more barren. transferable to other areas of arts and humanities that have medium- or long-term team-based collaborations. In this paper, we apply modern quantitative analysis meth- Index Terms—Popular music, network analysis, success pre- ods, including statistical analysis, social network analysis, and diction. machine learning, to a collection of 4,600 music groups and bands. We build a network of groups, quantify groups’ success, I. Introduction correlate it with network measures, and attempt to predict Exploring and predicting the success of creative collabo- success, based solely on the network measures.
    [Show full text]
  • The Balkans of the Balkans: the Meaning of Autobalkanism in Regional Popular Music
    arts Article The Balkans of the Balkans: The Meaning of Autobalkanism in Regional Popular Music Marija Dumni´cVilotijevi´c Institute of Musicology, Serbian Academy of Sciences and Arts, 11000 Belgrade, Serbia; [email protected] Received: 1 April 2020; Accepted: 1 June 2020; Published: 16 June 2020 Abstract: In this article, I discuss the use of the term “Balkan” in the regional popular music. In this context, Balkan popular music is contemporary popular folk music produced in the countries of the Balkans and intended for the Balkan markets (specifically, the people in the Western Balkans and diaspora communities). After the global success of “Balkan music” in the world music scene, this term influenced the cultures in the Balkans itself; however, interestingly, in the Balkans themselves “Balkan music” does not only refer to the musical characteristics of this genre—namely, it can also be applied music that derives from the genre of the “newly-composed folk music”, which is well known in the Western Balkans. The most important legacy of “Balkan” world music is the discourse on Balkan stereotypes, hence this article will reveal new aspects of autobalkanism in music. This research starts from several questions: where is “the Balkans” which is mentioned in these songs actually situated; what is the meaning of the term “Balkan” used for the audience from the Balkans; and, what are musical characteristics of the genre called trepfolk? Special focus will be on the post-Yugoslav market in the twenty-first century, with particular examples in Serbian language (as well as Bosnian and Croatian). Keywords: Balkan; popular folk music; trepfolk; autobalkanism 1.
    [Show full text]
  • Punk Aesthetics in Independent "New Folk", 1990-2008
    PUNK AESTHETICS IN INDEPENDENT "NEW FOLK", 1990-2008 John Encarnacao Student No. 10388041 Master of Arts in Humanities and Social Sciences University of Technology, Sydney 2009 ii Acknowledgements I would like to thank my supervisor Tony Mitchell for his suggestions for reading towards this thesis (particularly for pointing me towards Webb) and for his reading of, and feedback on, various drafts and nascent versions presented at conferences. Collin Chua was also very helpful during a period when Tony was on leave; thank you, Collin. Tony Mitchell and Kim Poole read the final draft of the thesis and provided some valuable and timely feedback. Cheers. Ian Collinson, Michelle Phillipov and Diana Springford each recommended readings; Zac Dadic sent some hard to find recordings to me from interstate; Andrew Khedoori offered me a show at 2SER-FM, where I learnt about some of the artists in this study, and where I had the good fortune to interview Dawn McCarthy; and Brendan Smyly and Diana Blom are valued colleagues of mine at University of Western Sydney who have consistently been up for robust discussions of research matters. Many thanks to you all. My friend Stephen Creswell’s amazing record collection has been readily available to me and has proved an invaluable resource. A hearty thanks! And most significant has been the support of my partner Zoë. Thanks and love to you for the many ways you helped to create a space where this research might take place. John Encarnacao 18 March 2009 iii Table of Contents Abstract vi I: Introduction 1 Frames
    [Show full text]
  • The Wiki Music Dataset: a Tool for Computational Analysis of Popular Music
    The Wiki Music dataset: A tool for computational analysis of popular music Fabio Celli Profilio Company s.r.l. via sommarive 18, 38123 Trento, Italy Email: fabio@profilio.co Abstract—Is it possible use algorithms to find trends in monic and timbral properties that brought changes in music the history of popular music? And is it possible to predict sound around 1964, 1983 and 1991 [14]. Beside these research the characteristics of future music genres? In order to answer fields, there is a trend in the psychology of music that studies these questions, we produced a hand-crafted dataset with the how the musical preferences are reflected in the dimensions intent to put together features about style, psychology, sociology of personality [11]. From this kind of research emerged the and typology, annotated by music genre and indexed by time MUSIC model [20], which found that genre preferences can and decade. We collected a list of popular genres by decade from Wikipedia and scored music genres based on Wikipedia be decomposed into five factors: Mellow (relaxed, slow, and ro- descriptions. Using statistical and machine learning techniques, mantic), Unpretentious, (easy, soft, well-known), Sophisticated we find trends in the musical preferences and use time series (complex, intelligent or avant-garde), Intense (loud, aggressive, forecasting to evaluate the prediction of future music genres. and tense) and Contemporary (catchy, rhythmic or danceable). Is it possible to find trends in the characteristics of the genres? Keywords—Popular Music, Computational Music analysis, And is it possible to predict the characteristics of future genres? Wikipedia, Natural Language Processing, dataset To answer these questions, we produced a hand-crafted dataset with the intent to put together MUSIC, style and sonic features, I.
    [Show full text]
  • Quantifying Music Trends and Facts Using Editorial Metadata from the Discogs Database
    QUANTIFYING MUSIC TRENDS AND FACTS USING EDITORIAL METADATA FROM THE DISCOGS DATABASE Dmitry Bogdanov, Xavier Serra Music Technology Group, Universitat Pompeu Fabra [email protected], [email protected] ABSTRACT Discogs metadata contains information about music re- leases (such as albums or EPs) including artists name, track While a vast amount of editorial metadata is being actively list including track durations, genre and style, format (e.g., gathered and used by music collectors and enthusiasts, it vinyl or CD), year and country of release. It also con- is often neglected by music information retrieval and mu- tains information about artist roles and relations as well sicology researchers. In this paper we propose to explore as recording companies and labels. The quality of the data Discogs, one of the largest databases of such data available in Discogs is considered to be high among music collec- in the public domain. Our main goal is to show how large- tors because of its strict guidelines, moderation system and scale analysis of its editorial metadata can raise questions a large community of involved enthusiasts. The database and serve as a tool for musicological research on a number contains contributions by more than 347,000 people. It of example studies. The metadata that we use describes contains 8.4 million releases by 5 million artists covering a music releases, such as albums or EPs. It includes infor- wide range of genres and styles (although the database was mation about artists, tracks and their durations, genre and initially focused on Electronic music). style, format (such as vinyl, CD, or digital files), year and Remarkably, it is the largest open database containing country of each release.
    [Show full text]
  • Sooloos Collections: Advanced Guide
    Sooloos Collections: Advanced Guide Sooloos Collectiions: Advanced Guide Contents Introduction ...........................................................................................................................................................3 Organising and Using a Sooloos Collection ...........................................................................................................4 Working with Sets ..................................................................................................................................................5 Organising through Naming ..................................................................................................................................7 Album Detail ....................................................................................................................................................... 11 Finding Content .................................................................................................................................................. 12 Explore ............................................................................................................................................................ 12 Search ............................................................................................................................................................. 14 Focus ..............................................................................................................................................................
    [Show full text]
  • Dive Covenant Ah Cama-Sotz Revolting Cocks
    edition April - June 2017 free of charge, not for sale 25 quarterly published music magazine DIVE COVENANT AH CAMA-SOTZ REVOLTING COCKS ARSENIC OF JABIR | FROZEN NATION | ALVAR CAUSENATION | DOLLS OF PAIN | VUDUVOX BESTIAL MOUTHS | 31ST TILE | ALPHAMAY X-MOUTH SYNDROME | LLUMEN | ELM - 1 - www.peek-a-boo-magazine.be - 2 - contents 04 CD reviews 22 Interview REVOLTING COCKS 06 Interview ALPHAMAY 24 Interview VUDUVOX 08 Interview DOLLS OF PAIN 26 CD reviews 10 Interview LLUMEN 28 Interview COVENANT 12 Interview ELM 30 CD reviews 14 Interview X-MOUTH SYNDROME 32 Interview AH CAMA-SOTZ 16 Interview ALVAR 18 Interview DIVE 34 Interview ARSENIC OF JABIR 20 Interview BESTIAL MOUTHS 35 Calendar Peek-A-Boo Magazine • BodyBeats Productions • Tabaksvest 40 • 2000 Antwerpen • contact@ and [email protected] colophon ORGANISATION EDITORS WRITERS (continued) BODYBEATS Productions Gea STAPELVOORT Paul PLEDGER www.bodybeats.be Leanne AITKEN Ron SCHOONWATER Dimitri CAUVEREN Sara VANNACCI Stef COLDHEART Wool-E Shop Tine SWAENEPOEL Ward DE PRINS Dries HAESELDONCKX Wim LENAERTS Bunkerleute WRITERS Xavier KRUTH Frédéric COTTON Charles “Chuck” MOORHOUSE Le Fantastique Fred GADGET PHOTOGRAPHERS PARTNERS Gustavo A. ROSELINSKY Benny SERNEELS Dark Entries team Jurgen BRAECKEVELT Marquis(pi)X www.darkentries.be Manu L DASH Gothville team Mark VAN MULLEM MAGAZINE & WEBSITE www.gothville.com Masha KASHA Ward DE PRINS - 3 - www.peek-a-boo-magazine.be ONCEHUMAN - Evolution (CD) (Ear Music) Sometimes you have to take a leap of faith when choosing a CD for your listening pleasure and experience. However, mostly, you take a diligent peek at the visual aspects (artwork, layout, photography...).
    [Show full text]
  • Table of Contents
    1 •••I I Table of Contents Freebies! 3 Rock 55 New Spring Titles 3 R&B it Rap * Dance 59 Women's Spirituality * New Age 12 Gospel 60 Recovery 24 Blues 61 Women's Music *• Feminist Music 25 Jazz 62 Comedy 37 Classical 63 Ladyslipper Top 40 37 Spoken 65 African 38 Babyslipper Catalog 66 Arabic * Middle Eastern 39 "Mehn's Music' 70 Asian 39 Videos 72 Celtic * British Isles 40 Kids'Videos 76 European 43 Songbooks, Posters 77 Latin American _ 43 Jewelry, Books 78 Native American 44 Cards, T-Shirts 80 Jewish 46 Ordering Information 84 Reggae 47 Donor Discount Club 84 Country 48 Order Blank 85 Folk * Traditional 49 Artist Index 86 Art exhibit at Horace Williams House spurs bride to change reception plans By Jennifer Brett FROM OUR "CONTROVERSIAL- SUffWriter COVER ARTIST, When Julie Wyne became engaged, she and her fiance planned to hold (heir SUDIE RAKUSIN wedding reception at the historic Horace Williams House on Rosemary Street. The Sabbats Series Notecards sOk But a controversial art exhibit dis­ A spectacular set of 8 color notecards^^ played in the house prompted Wyne to reproductions of original oil paintings by Sudie change her plans and move the Feb. IS Rakusin. Each personifies one Sabbat and holds the reception to the Siena Hotel. symbols, phase of the moon, the feeling of the season, The exhibit, by Hillsborough artist what is growing and being harvested...against a Sudie Rakusin, includes paintings of background color of the corresponding chakra. The 8 scantily clad and bare-breasted women. Sabbats are Winter Solstice, Candelmas, Spring "I have no problem with the gallery Equinox, Beltane/May Eve, Summer Solstice, showing the paintings," Wyne told The Lammas, Autumn Equinox, and Hallomas.
    [Show full text]
  • Why Do I Like This Song? and Other Important Questions GREG STIFFLER COMMUNITY COLLEGE of BALTIMORE COUNTY [email protected] Can’T Get It Outta’ My Head
    Why Do I Like This Song? And Other Important Questions GREG STIFFLER COMMUNITY COLLEGE OF BALTIMORE COUNTY [email protected] Can’t Get it Outta’ My Head What songs will you always remember? What songs give you chills or excite you? Selection 1: Selection 2: Selection 3: Selection 4: Selection 5: Selection 6: Born This Way From an early age, form opinions on music and sound From as early as age 5, we can decipher changes in tempo, happy/sad excerpts by extension (study 9) Major chords typically envoke happiness, elation Minor chords typically envoke sadness, stoicism or sad feelings (study 9) Emotional associations established early in development (study 14) Have you ever anticipated a refrain or hook in a song? This comes from one place… Check My Brain Important parts of the brain Ventral striatum Nucleus accumbens (NAc) Ventral tegmental area (VTA) Hippocampus Amygdala Insular cortex (insula) Periaqueductal gray (PAG)/pedunculopontine tegmental nucleus (PPT) Pleasure center of the brain (study 13 and websites) Addiction center If I Only Had a Brain… But I don’t even like that song! Sex, Drugs, and Rock and Roll Doesn’t matter Food, sex increase dopamine activity in NAc (study 13) Only familiarity needed to stimulate amygdala, NAc (study8) Midbrain and reward center responses to music mimic those of cocaine Mere Exposure Effect studies; NAc, VTA, and insula area “…mere repetition of melodies is sufficient in increase affective Research suggest stimulation of PAG responses to these melodies.” and PPT
    [Show full text]
  • Text-Based Description of Music for Indexing, Retrieval, and Browsing
    JOHANNES KEPLER UNIVERSITAT¨ LINZ JKU Technisch-Naturwissenschaftliche Fakult¨at Text-Based Description of Music for Indexing, Retrieval, and Browsing DISSERTATION zur Erlangung des akademischen Grades Doktor im Doktoratsstudium der Technischen Wissenschaften Eingereicht von: Dipl.-Ing. Peter Knees Angefertigt am: Institut f¨ur Computational Perception Beurteilung: Univ.Prof. Dipl.-Ing. Dr. Gerhard Widmer (Betreuung) Ao.Univ.Prof. Dipl.-Ing. Dr. Andreas Rauber Linz, November 2010 ii Eidesstattliche Erkl¨arung Ich erkl¨are an Eides statt, dass ich die vorliegende Dissertation selbstst¨andig und ohne fremde Hilfe verfasst, andere als die angegebenen Quellen und Hilfsmittel nicht benutzt bzw. die w¨ortlich oder sinngem¨aß entnommenen Stellen als solche kenntlich gemacht habe. iii iv Kurzfassung Ziel der vorliegenden Dissertation ist die Entwicklung automatischer Methoden zur Extraktion von Deskriptoren aus dem Web, die mit Musikst¨ucken assoziiert wer- den k¨onnen. Die so gewonnenen Musikdeskriptoren erlauben die Indizierung um- fassender Musiksammlungen mithilfe vielf¨altiger Bezeichnungen und erm¨oglichen es, Musikst¨ucke auffindbar zu machen und Sammlungen zu explorieren. Die vorgestell- ten Techniken bedienen sich g¨angiger Web-Suchmaschinen um Texte zu finden, die in Beziehung zu den St¨ucken stehen. Aus diesen Texten werden Deskriptoren gewon- nen, die zum Einsatz kommen k¨onnen zur Beschriftung, um die Orientierung innerhalb von Musikinterfaces zu ver- • einfachen (speziell in einem ebenfalls vorgestellten dreidimensionalen Musik- interface), als Indizierungsschlagworte, die in Folge als Features in Retrieval-Systemen f¨ur • Musik dienen, die Abfragen bestehend aus beliebigem, beschreibendem Text verarbeiten k¨onnen, oder als Features in adaptiven Retrieval-Systemen, die versuchen, zielgerichtete • Vorschl¨age basierend auf dem Suchverhalten des Benutzers zu machen.
    [Show full text]