Environmental issues are increasingly discussed through social media applications. Consequently, researchers are beginning to question whether social media could represent a green virtual sphere: a virtual public space to discuss environmental issues not governed by a single authority in which anyone can access; however, limited empirical research has been conducted to date. In response, this study combines text analysis, social network analysis, and semi-structured interviews to determine whether discussions regarding the Alberta oil sands and the Northern Gateway Pipeline on Twitter – a micro-blogging site – reflect the characteristics of a green virtual sphere. It was found that Twitter is used to disseminate information, access news, and engage in debate, but there are limitations: not everyone has access to Twitter, the government may monitor online activity, and discussions appear to be dominated by environmentalists and environmental organizations. Twitter use on these issues only partially reflects the characteristics of a green virtual sphere.

People share a big part of their lives and opinions on platforms such as Facebook and Twitter. The companies behind these sites do their absolute best to collect as much data as possible. This data could be used to extract opinions in many different ways. Every company, organization or public person is probably curious on what is being said about them right now. There are also areas where opinions are related to the outcome of an event. Examples of such events are presidential elections or the Eurovision Song Contest. In these events, peoples’ votes will directly reflect the outcome of the elections or contests. We have developed a simplistic prototype that is able to predict the result of the Eurovision Song Contest using sentiment analysis on tweets. The prototype collects tweets about the event, performs sentiment analysis, and uses different filters to predict the ranks of the contestants. We evaluted our results with the actual voting results of the event and found a Pearson correlation of approximately 0.65. With more time and resources we believe that it is possible to create a highly accurate prediction model. It could be used in lots of different contexts. Politicians and their parties could use it to evaluate their campaigns. The press could use it to create more interesting news reports. Companies would be able to investigate their brand appreciation. A system like this could be used in many different fields.

Allt fler journalister världen över använder sig av mikrobloggen Twitter i sin yrkesutövning. Studier har visat att det handlar om allt från att hålla sig uppdaterad på nyhetsläget till att live-rapportera. Gruppen miljöjournalister valdes ut till den här studien vars syfte är att ta reda på mer om hur journalister använder Twitter och hur det påverkar deras arbete och inflöde. Miljöjournalister utmärker sig genom att de figurerar på ett brett och åsiktsspäckat fält där Twitter kan tänkas fylla en speciellt viktig funktion som plattform för kontakt med många olika aktörer. Fem miljöjournalister djupintervjuades för att få svar på frågorna. Utifrån bland annat Alfred Hermidas teorier kring ambient journalism kan miljöjournalisternas Twitter-användning förstås som ett sätt att känna av trender och få tidiga indikationer om vad som är aktuellt på deras område. Det fanns också antydningar om att Twitter breddar deras kontaktnät och tillgång till källor. På så sätt är det viktigt att se Twitter både ur ett makro- och mikroperspektiv där enskilda inlägg kan bidra med viktig kunskap. Att journalisterna ingick i ett nätverk på Twitter där åsiktsyttringar är vanligt tycktes inte ha påverkat deras syn på vikten av att förhålla sig objektiv, något som bland annat märktes på hur de föredrog att skicka vidare andras inlägg snarare än att skriva egna. Vidare gick det att skönja en pågående individualisering inom journalistiken. Delvis genom att Twitter är en plats där journalisterna lägger upp sitt egenproducerade material, men även med tanke på att miljöjournalisterna själva till viss del valde att följa andra journalister snarare än nyhetskanaler.

The concept of social media is top of mind for Canadians today. Decision makers, such as the Canadian government, try to identify ways in which they can benefit from tools such as Twitter. This thesis is an investigative analysis that studies how the Canadian government currently uses social media networks. Based on the conceptual framework of Rogers’ diffusion of innovations (2003) and Qualman’s socialnomics theory (2013), the study aims to define how members of the Parliamentary Press Gallery use the Government of Canada tweets. Through a series of interviews with members of the press and government communications executives, a themed analysis was conducted to demonstrate how Twitter is being used and if a dialogue exists between federal institutions and reporters. The research unveiled that the Government of Canada uses social media as a one-to-many broadcasting channel, not actively engaging in online dialogue with members of the press. Conversely, journalists use Twitter as a wire service to obtain instant information, and to gain insight from the consumers. However, they are not interested in conversing with the public service on social media as they want to protect the exclusivity of their story, and they criticize the anonymity of the government corporate accounts as having an impact on its credibility.

This experiment analyzes “tweets” gathered from Twitter and determines whether the positive or negative emotions conveyed in the contents of a massive collection of these tweets correlate with the percentage change in indexes of Standard & Poor's 500, the Dow Jones Industrial Average, and the NASDAQ stock markets. This experiment uses an algorithm that parses a random sample of live tweets and calculates their sentiment value, which is an indicator of whether a tweet is either negative or positive in its emotional content. The daily ­average sentiment value is then compared to the percent change in the stock exchanges.

Denna studie syftar till att ge en bild av varför konsumenter väljer att engagera sig i interaktionen genom företags kanaler på sociala medier. Det empiriska underlaget insamlades genom en kvantitativ studie i form av en enkätundersökning samt en kompletterande textanalys av inlägg skrivna på Twitter. Studien undersökte fyra bakomliggande processer till engagemang. Dessa engagemangsprocesser är socialisering, samvärdeskapande, lärande och word-of-mouth. Undersökningarna visar att socialisering och word-of-mouth är de största bakomliggande processerna till konsumenters engagemang i jämförelse med lärande och samvärdeskapande. Enligt den kvantitativa undersökningen har word-of-mouth störst inverkan på engagemang medan resultatet av textanalysen visar socialisering som det största motivet för konsumenters engagemang. Efter utvärdering av de båda undersökningarna och dess resultat dras slutsatsen att konsumenter väljer att engagera sig i interaktionen genom socialisering.

Σε αυτήν την εργασία χρησιμοποιούμε το μέσο κοινωνικής δικτύωσης "twitter" (https://twitter.com/) για την συλλογή μηνυμάτων που αφορούν τις εξελίξεις στην ευρωζώνη και την εφαρμογή μεθόδων επιβλεπόμενης μηχανικής μάθησης για την "εκπαίδευση" ενός κατηγοριοποιητή ο οποίος θα διαχωρίζει τα μηνύματα σε "θετικά" και "αρνητικά" ανάλογα με την είδηση ή την άποψη που περιέχουν. Οι μέθοδοι κατηγοριοποίησης που εφαρμόστηκαν ήταν οι k πλησιέστεροι γείτονες, μηχανές διανυσμάτων υποστήριξης και αφελής Μπεϊζιανός κατηγοριοποιητής. Ο ταξινομητής θα μπορούσε να χρησιμοποιηθεί σε ένα απλό πρόγραμμα το οποίο ημερησίως θα συλλέγει και θα ταξινομεί, αυτομάτως, σχετικά μηνύματα. Μία μακρυπρόθεσμη χρήση ενός τέτοιου προγράμματος θα μας έδινε σαν αποτέλεσμα δεδομένα σε μορφή χρονοσειράς τα οποία στην συνέχεια θα μπορούσαν να αναλυθούν για την εξαγωγή, πιθανώς, χρήσιμων συμπερασμάτων. / --

Submitted by Pós-Com Pós-Com (pos-com@ufba.br) on 2018-05-11T12:23:44Z No. of bitstreams: 1 DISSERTAÇAO _MARTA_FLORENCIA_GOLDSMAN_2018 (1).pdf: 11102696 bytes, checksum: 7e709ca12213c20b844fe039baebe9fd (MD5) / Rejected by Vania Magalhaes (magal@ufba.br), reason: on 2018-05-11T12:33:56Z (GMT) / Submitted by Pós-Com Pós-Com (pos-com@ufba.br) on 2018-05-11T13:07:02Z No. of bitstreams: 1 DISSERTAÇAO _MARTA_FLORENCIA_GOLDSMAN_2018 (1).pdf: 11102696 bytes, checksum: 7e709ca12213c20b844fe039baebe9fd (MD5) / Approved for entry into archive by Setor de Periódicos (per_macedocosta@ufba.br) on 2018-05-11T15:47:19Z (GMT) No. of bitstreams: 1 DISSERTAÇAO _MARTA_FLORENCIA_GOLDSMAN_2018 (1).pdf: 11102696 bytes, checksum: 7e709ca12213c20b844fe039baebe9fd (MD5) / Made available in DSpace on 2018-05-11T15:47:19Z (GMT). No. of bitstreams: 1 DISSERTAÇAO _MARTA_FLORENCIA_GOLDSMAN_2018 (1).pdf: 11102696 bytes, checksum: 7e709ca12213c20b844fe039baebe9fd (MD5) / FAPESB / Entre abril e agosto de 2016, uma onda de protestos tomou conta das ruas e das redes sociais digitais na Argentina. Pela primeira vez, um protesto massivo contra a criminalização do aborto chegou a ser “tendência” no Twitter. Os movimentos de mulheres e feministas organizados acompanhados por sindicatos, representantes de partidos políticos e figuras públicas se posicionaram através da hashtag #LibertadParaBelen, que exigia a liberação de uma jovem presa durante dois anos por causa de um aborto espontâneo. Neste trabalho, analisamos a rede formada no Twitter por essa manifestação, que debateu o direito das mulheres decidirem sobre o aborto. Realizamos uma raspagem de 12.050 tweets, capturados entre 25 de abril e 02 de novembro de 2016. Com esses rastros digitais, geramos um grafo em Gephi que busca representar a visualização e identificação dos atores nesta rede de interações durante esse recorte de tempo. Também criamos un Mandala semântica (oferecida pelo LABIC – UFES) que permite identificar os termos mais frequentes nas conversações registradas no Twitter. Nos concentramos em estudar o Twitter como uma plataforma com uma política de dados e uma série de réguas específicas que definem a interação. A investigação empregou uma abordagem fundamentada nos métodos de pesquisa digital (CHARMAZ, Kathy, 2009; FRAGOSO, Suely; RECUERO, Raquel; AMARAL, Adriana, 2011; BITTENCOURT, Maíra, 2017) que nos permitiu elaborar procedimentos própios para a análise dos dados coletados. Nossos resultados mostram, de maneira complexa, o amadurecimento de um debate histórico sobre o direito ao aborto e a busca por sua despenalização total. / Between April and August 2016, a wave of protests took place in the streets and in the digital social networks in Argentina. For the first time in years, a massive protest linked to the criminalization of women who abort became a "trend" on Twitter. The movements of women and feminists organized together with unions, representatives of political parties and publicly recognized figures positioned themselves through the hashtag #LibertadParaBelen demanding the release of a young women imprisoned for two years because of an spontaneous abortion. In this work we analyze the network formed by the manifestation that speaks about the right to decide of women on Twitter. We do this by scraping 12,050 tweets captured from April 25 to November 2, 2016. With these digital traces we generated a graph in Gephi which aim is to represent the visualization and identification of the actors in this network of interactions as long of that time lapse. We also created a semantic mandala (offered by the LABIC - UFES) that allowed us to identify the most frequent terms in the conversations that took place on Twitter. We concentrate on studying Twitter as a platform with a data policy and a series of specific rules that define the interaction. The research used an approach based on digital research methods (CHARMAZ, Kathy, 2009, FRAGOSO, Suely, RECUERO, Raquel, AMARAL, Adriana, 2011, BITTENCOURT, Maira, 2017) that allowed us to develop our own procedures for data analysis. Our results show, in a complex way, the maturation of an historical debate on the right to abortion in the advance towards total decriminalization.

Studiens syfte är att undersöka om uppmätt sentiment på Twitter kan vara en förutsägande faktor för priset på Bitcoin. En kvantitativ undersökning genomförs med regressionsmodeller där data inhämtas från Twitter i realtid. Resultatet indikerar ett svagt samband där bäst resultat erhölls med en tidsfördröjning av sentiment på 16 timmar, vilket tyder på att det kan finnas möjligheter att använda Twitter för att förutspå förändringar av priset på Bitcoin. Variationen av resultat för olika tidsperioder gör dock att det är svårt att dra generella slutsatser av studien.

[pt] Nesta dissertação, apresentamos um processador linguístico que resolve a tarefa de Anotação morfossintática de mensagens em português postadas no Twitter. Ao analisar as mensagens escritas por brasileiros no Twitter, é fácil verificar que novos caracteres são introduzidos no alfabeto e também que novas palavras são adicionadas ao idioma. Além disso, observamos que essas mensagens são sintaticamente mal formadas. Isto impossibilita o uso nessas mensagens de diversos processadores linguísticos existentes para o português. Resolvemos esse problema considerando essas mensagens como escritas em uma nova língua, o português-twitter. O alfabeto dessa nova língua contém o alfabeto do português e o seu vocabulário contém o vocabulário da língua portuguesa. Porém, suas gramáticas são diferentes. Para construir os processadores desta nova linguagem, utilizamos a técnica de aprendizado supervisionado denominada Entropy Guided Transformation Learning (ETL). Adicionalmente, para treinar os processadores ETL, construímos um corpus anotado de mensagens em português-twitter. Não temos conhecimento da existência de outros Anotadores Morfossintáticos para o português-twitter. Porém, sabemos que, no estado-da-arte da Anotação Morfossintática para o português, a acurácia é de aproximadamente 96%, variando de acordo com o conjunto de classes escolhido. Construímos o processador composto de dois estágios, um morfológico e um contextual. Como métrica de avaliação, adotamos a acurácia, que mede quantos por cento do corpus foi anotado corretamente. Nossos resultados experimentais apresentam uma acurácia de 90,24% para o anotador proposto. Isto corresponde a um aprendizado significativo, pois o sistema inicial tem uma acurácia de apenas 76,58%. Este resultado é compatível com o aprendizado observado nos correspondentes processadores na língua portuguesa. / [en] In this paper we present a language processor that solves the task of Morphosyntactic Tagging of messages posted in Portuguese on Twitter. By analyzing the messages written by Brazilian on Twitter, it is easy to notice that new characters are introduced in the alphabet and also that new words are added to the language. Furthermore, we note that these messages are syntactically malformed. This precludes the use of existing Portuguese processors in these messages, nevertheless this problem can be solved by considering these messages as written in a new language, the Portuguese-Twitter. Both the alphabet and the vocabulary of such idiom contain features of Portuguese. However, the grammar is are different. In order to build the processors for this new language, we have used a supervised learning technique known as Entropy Guided Transformation Learning (ETL). Additionally, to train ETL processors, we have built an annotated corpus of messages in Portuguese-Twitter. We are not aware of any other taggers for the Morphosyntactic Portuguese-Twitter task, thus we have compared our tagger to the the accuracy of state-of-art Morphosyntactic Annotation for Portuguese, which has accuracy around 96% depending on the tag set chosen. To assess the quality of the processor, we have used accuracy, which measures how many tokens were tagged correctly. Our experimental results show an accuracy of 90,24% for the proposed Morphosyntatic Tagger. This corresponds to significant learning, since the initial baseline system has an accuracy of only 76,58%. This finding is consistent with the observed learning for the corresponding regular Portuguese taggers.