Musical networks : the case for a neural network methodology in advertisement music research

Olivier, Hannelore (2005-04)

Thesis (M.Mus.)--University of Stellenbosch, 2005.

Thesis

ENGLISH ABSTRACT: Countless scientists had been struggling for centuries to find a significant connection between cognition, emotion and reasoning – resulting in today’s rather embarrassingly imperfect understanding of even the most basic human cognition. We should apprehend that it is unlikely that major breakthroughs in the Cognitive Sciences, Psychology, Sociology or the Medical Sciences will elucidate everything about the human brain and -behaviour in the very near future. Realizing this, it is realistic that we should transfer our attention to things that we do know and understand, and reconsider the power that lies in the integration of results and an interdisciplinary perspective in research. Using the tools we have to our disposal today – digital tools such as ANNs which did not exist a few decades before – this is actually readily viable today. This thesis demonstrates that it is possible to break the traditional boundaries that have periodically prevented the Humanities and the Natural Sciences to join forces towards a greater understanding of human beings. By using ANNs, we are able to merge data from any subfield within the Humanities and Natural Sciences in a single study. The results, interpretations and applications which could develop from such a study would certainly be more inclusive than those derived from research conducted in one or two of these fields in isolation. Sufficient evidence is provided in this dissertation to support a methodology which employs an artificial neural network to assist with decision-making processes related to the choice of advertisement music. The main objective of this endeavour is to establish the feasibility of combining data from many diverse fields, in the creation of an ANN that can be helpful in research regarding South African advertisement music. The thesis explores the notion that knowledge from many interdisciplinary study fields ought to play a leading role in the creation and assessment of effective, target-group-specific advertisement music. In obtaining this goal, it examines the probability of producing a computer-based tool which can assist people working in the advertising industry to obtain an educated match between product, consumer, and advertisement music. Taking a multidisciplinary point of view, the author suggests a methodology for the design of a digital tool in the form of a musical network model. It is concluded that, by using this musical network, it is indeed possible to guarantee a functional musically-paired commercial, which effectively addresses its target-group and has an appropriate emotional effect in support of the marketing goals of the advertising agent. The thesis also demonstrates that it is possible to gain new insights regarding a fairly unstudied discipline, without necessarily conducting new research studies in the specified field. The thesis proves that - by taking an interdisciplinary approach and by using ANNs - it is possible to attain new data that is scientifically valid, even in an unacknowledged field such as South African advertisement music. Although the scope of the thesis does not provide for the actual implementation of the musical network, the feasibility of the conceptual idea is thoroughly examined, and it is concluded that the theory in it’s entirely is definitely feasible, and can be implemented in a future study.

AFRIKAANSE OPSOMMING: Vir eeue al probeer wetenskaplikes ‘n betekenisvolle verwantskap tussen denke, emosie en redenasie vind. Nietemin het ons vandag slegs ‘n beperkte begrip van selfs die mees basiese menslike kognisie. Ons moet besef dat dit onwaarskynlik is dat deurbrake in die Kognitiewe Wetenskappe, Sielkunde, Sosiologie of die Mediese Wetenskap in die nabye toekoms die volle funksionaliteit van die menslike brein en gedrag sal bekendmaak. Met inagname hiervan, is ‘n aandagsverskuiwing geoorloof - na die dinge wat ons wel weet en verstaan. Die enorme potensiaal opgesluit in die integrasie van resultate en ‘n interdissiplinêre navorsingsperspektief behoort gevolglik heroorweeg te word. Ons beskik tans oor meer as voldoende digitale hulpbronne, waaronder kunsmatige neurale netwerke, wat wel so ‘n benadering kan bewerkstellig. In hierdie tesis word daar gedemonstreer dat dit moontlik is om die grense wat tradisioneel ‘n samewerking tussen die Geestes- en Natuurwetenskappe beperk het, af te breek - ‘n werkswyse wat noodwendig sal lei tot ‘n beter begrip van die mens. Kunsmatige neurale netwerke maak dit moontlik om navorsingsdata uit die Geestes- en Natuurwetenskappe te kombineer in ‘n enkele onderneming. Die bevindinge, interpretasies en toepasings wat potensieel uit so ‘n metodologie sou kon voortspruit, is sonder twyfel meer omvattend as dié afkomstig vanuit ‘n eendimensionele studie. Voldoende bewyse word deur die loop van hierdie studie voorgehou ter ondersteuning van ‘n kunsmatige neurale netwerk-metodologie in die assistering van besluitnemingsprosesse rakende advertensiemusiek. Die hoofdoelwit van die onderneming is om te toets of die ontwerp van ‘n kunsmatige neurale netwerk - deur die kombinasie van data uit diverse studierigtings - wel geoorloof en funksioneel sou kon wees. Die aanname dat inligting uit ‘n aantal interdissiplinêre studierigtings ‘n prominente rol behoort te speel tydens die skep en beoordeling van effektiewe, teikengroep-gerigte advertensiemusiek, word gevolglik ondersoek. Om hierdie objektief te bewerkstellig, word die waarskynlikheid bestudeer na die ontwerp van ‘n rekenaargebaseerde hulpbron - wat mense in die advertensiewese behulpsaam kan wees om ‘n berekende en ingeligte keuse uit te oefen om produk, verbruiker en advertensiemusiek te laat pas. Die outeur benader die probleem vanuit ‘n multidissiplinêne oogpunt, en stel ‘n werkswyse voor vir die ontwerp van ‘n digitale hulpbron – in die vorm van ‘n musikale netwerk model. Daar word bevind dat - deur die gebruik van die voorgestelde model, dit wel moontlik is om die funksionaliteit van ‘n musiekgepaarde advertensie te verseker. Verder word daar gedemonstreer dat nuwe insigte rakende ‘n grotendeels afgeskeepte studierigting soos Suid-Afrikaanse advertensiemusiek moeiteloos bekom kan word, sonder om noodwendig navorsing binne die spesifieke gebied te loods. Laasgenoemde is doenbaar deur ‘n interdissiplinêre navorsingsbenadering, gekombineerd met ‘n kunsmatige neurale netwerk-metodologie. Die omvang van hierdie studie maak nie voorsiening vir die implementering van die musikale netwerk nie. Nietemin word die werkbaarheid van die konseptuele idee in diepte ondersoek, met die gevolgtrekking dat die teorie in sy geheel sonder twyfel prakties is, en in ‘n toekomstige studie geïmplementeer kan word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/16618
This item appears in the following collections: