Thesis (MA)--Stellenbosch University, 2013. / ENGLISH ABSTRACT: The constant growth in information and communications technology (ICT) broadens the capabilities of researchers to understand and interpret the social world. New multidisciplinary methods of measuring social realities are constantly being developed and improved. This methodological study aims to incorporate novel methods of automated, computerised content- and sentiment analysis software in order to produce a usable instrument for measuring public expressions on democracy from the social network site, Twitter. Adapting methods developed by O‟Connor et al. (2010) and using an online platform called Chatterbox Analytics Ltd., this study attempts to complement traditional survey research data on democracy. The creation of this method allows researchers to automatically and without deep programming knowledge, extract and analyse opinionated data from a substantial segment of the population, namely Twitter users. This is the first study, as far as this researcher is aware, that develops automated instruments for the measurement of expressions on democracy from online social networks.
The rationale for the development of this instrument lies in the apparent recent rise in negative sentiment on democracy in academic spheres and political realities alike. Throughout history support for democracy has swayed, which in turn placed democracy on a rocky path of development. As negative opinions on democracy became overwhelming this system of governance was forced to adapt by changing its core structure. This could be the case again today. Authors such as Crozier, Huntington and Watunaki (1975) have already warned of a crisis of democracy in the 1970s. More recently authors such as Dalton (2004) and van Beek (2012) expressed opinions indicating that globally citizens are steadily becoming more critical of political parties, politicians, political policies and the whole democratic process. These negative sentiments might be indicative of what van Beek (2012:12) calls “an erosion in support for democracy”.
Traditional survey projects examining opinions on democracy, such as the World Values Survey, offer insights into global expressions on democracy, but are time consuming, costly and complicated to conduct. The method developed in this study allows researchers to complement traditional survey data with insights that are automatically gathered and analysed from the influential social media network, Twitter.
This introductory study finds that automated analysis of expressions on democracy from Twitter is indeed feasible. In order to express the extent to which the developed method is feasible, this study offers an example case, examining expressions on democracy from Twitter for the period of 1 May to 31 July 2012. Furthermore the study offers a quantitative manual evaluation of the accuracy of the developed measurement instrument. With the present level of content- and sentiment analysis technology this study finds that the accuracy of the results from this method, though informative, is still limited. The study therefore concludes with an advisory section highlighting methods for future studies to improve on the accuracy of this measurement instrument. / AFRIKAANSE OPSOMMING: Deur die konstante groei in inligting- en kommunikasietegnologie (IKT) word die veld waaruit navorsers kan put om die sosiale wêreld te verstaan en te interpreteer al groter. Nuwe multidissiplinêre metingsmetodes om die sosiale werklikheid te verstaan, word konstant ontwikkel en verbeter. Hierdie metodologiese studie beoog om nuwe metodes van geoutomatiseerde, gerekenaariseerde, inhouds- en sentimentanalisesagteware saam te voeg om 'n bruikbare instrument vir die meting van publieke uitsprake oor demokrasie, op die sosiale netwerk Twitter te skep. Metodes wat deur O'Connor et al. (2010) en die aanlyn platform, Chatterbox Analytics Ltd ontwikkel is, is aangepas om 'n instrument te skep wat gebruik kan word om by te dra tot tradisionele opname-navorsingsdata oor demokrasie. Hierdie nuwe metode sal navorsers toelaat om outomaties en sonder veel programmeringsvaardighede opiniegelaaide data van 'n beduidende segment van die samelewing, naamlik Twitter gebruikers, te verkry en te analiseer. Hierdie is die eerste studie, sover hierdie navorser bewus is, wat geoutomatiseerde instrumente ontwikkel vir die meting van uitsprake oor demokrasie op sosiale netwerke.
Die beweegrede vir die ontwikkeling van hierdie instrument vloei uit die oënskynlike onlangse styging in negatiewe sentiment oor demokrasie in sowel akademiese- as politieke sirkels. Die ondersteuning vir demokrasie het dwarsdeur die geskiedenis sy hoogte en laagtepunte gehad en hierdie onsekerheid het die ontwikkelingspad van demokrasie taamlik met dorings besaai. Die hele kernstruktuur van híérdie regeringsisteem was geforseer om hewig aan te pas elke keer wanneer die negatiewe opinies oor demokrasie oorweldigend geraak het. Dit wil voorkom asof demokrasie weer op die rand van verandering is. Skrywers soos Crozier, Huntington en Watunaki (1975) het alreeds in die 1970's gewaarsku teen 'n krisis rondom demokrasie. Meer onlangs het skrywers soos Dalton (2004) en van Beek (2012) opinies gelug dat burgers reg oor die wêreld stadig maar seker besig is om meer krities op politieke partye, politici en oor politieke beleide en die demokratiese proses in geheel raak. Hierdie negatiewe sentimente mag dalk 'n aanduiding wees van “'n erosie in die ondersteuning van demokrasie,”, soos Van Beek (2012:12) dit noem.
Tradisionele opname-navorsingsprojekte wat na opinies oor demokrasie kyk, soos die „World Values Survey‟, bied wel 'n blik op globale uitsprake oor demokrasie, maar is duur, tydsaam en moeilik om deur te voer. Die metode wat in hierdie studie ontwikkel is, kan bykomende insigte tot tradisionele opname-navorsingsprojekte bring. Díé insigte word outomaties ingesamel en geanaliseer vanuit die invloedryke sosiale netwerk, Twitter.
Hierdie inleidende studie bevind dat geoutomatiseerde analise van uitsprake oor demokrasie op Twittter inderdaad moontlik is. Ten einde die mate van sukses van hierdie metode wat ontwikkel is te illustreer, analiseer hierdie studie 'n datagreep van uitsprake oor demokrasie uit Twitter oor die tydperk 1 Mei tot 31 Julie 2012. Hierbenewens bied die studie ook 'n kwantitatiewe waardasie per hand wat die akkuraatheid van die ontwikkelde meetingsinstrument toets. Met die huidige ontwikkelingsvlak van inhouds- en sentiment-analisetegnologie vind hierdie studie dat die akkuraatheid van hierdie metode, alhoewel informatief is, tog beperk is. Die studie sluit af met 'n afdeling wat advies oor hoe verdere studies die akkuraatheid van hierdie meetinstrument kan verbeter, toelig.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:sun/oai:scholar.sun.ac.za:10019.1/79986 |
| Date | 03 1900 |
| Creators | Lutz, Barend Frederik |
| Contributors | Du Toit, P. V. D. P., Stellenbosch University. Faculty of Arts and Social Sciences. Dept. of Political Science. |
| Publisher | Stellenbosch : Stellenbosch University |
| Source Sets | South African National ETD Portal |
| Language | en_ZA |
| Detected Language | English |
| Type | Thesis |
| Format | xi, 109 p. Ill. |
| Rights | Stellenbosch University |
Page generated in 0.0033 seconds