Suchergebnisse

This is a conference paper. ; The effective management and efficient use of data is a major challenge for government, academia and industry. The scale, diversity and distributed nature of current and emerging data assets are increasing. For example, every sixty seconds there are 433,000 tweets, 2.66 million Google searches, 4.7 million Tumblr posts, and 139 million emails sent. As data becomes ever more ubiquitous and critical to decision making it is vital that it is turned into meaningful information. The aim of the research discussed in this paper was to determine the usefulness of the EMOTIVE system, developed by the authors, to track the Scottish Independence Referendum and if it could be used to track the general public emotions towards the UK 2015 General Election by creating a new mobile phone application called VOTEBEE (Voter Opinion Tracked by EMOTIVE Engine). In this paper we employ an ontology engineering approach to the problem of fine-grained emotion detection in sparse messages. Messages are also processed using a custom NLP pipeline, which is appropriate for the sparse and informal nature of text encountered on micro-blogs or tweets. Our approach detects a range of eight high-level emotions; anger, confusion, disgust, fear, happiness, sadness, shame and surprise. The paper analyses the Scottish Independence Referendum data using EMOTIVE to determine what can we learn from monitoring social media networks and if there is a requirement to develop systems like VOTEBEE to help determine voter interest and increase debates in the lead up to the 2015 UK General Election.

This is a conference paper. ; We explore President Donald Trump's tweeting habits and the effects they have on his followers and the American media. To gain a comprehensive picture of Trump's tweeting habits as President of the United States, we have undertaken a study of Trump's personal @realDonaldTrump Twitter account, focusing on his campaign, the transition period before his presidency, and first 200 days in office. We employ three state-ofthe-art computational tools to analyse sentiment, emotions and psycholinguistic features in Trump's tweets to decipher what it is about his communication methods that generates the highest responses and retweets. We find that during the first 200 days of presidency, an accusative tone of discourse was most frequently used by @realDonaldTrump, and among a number of significant emotional patterns, we observe an intriguing correlation in that the more negative the overall message of a tweet is, the more likely it is to be re-tweeted, favourited, and discussed in American mainstream media. We also used data from the recently released PTDC corpus (Political Twitter Discourse Corpus) consisting of 205,303 original tweets of all current US state governors, members of the US Senate, and members of Congress and found that Trump's tweeting style is significantly different along a central dimension of language on Twitter. Our findings suggest that the general public on Twitter respond more actively to negative and less analytic language, and in turn, the language on Twitter directed at Trump is highly emotional and often contains negative sentiment.

This is a conference paper. ; With the uptake of social media, such as Facebook and Twitter, there is now a vast amount of new user generated content on a daily basis, much of it in the form of short, informal free-form text. Businesses, institutions, governments and law enforcement organisations are now actively seeking ways to monitor and more generally analyse public response to various events, products and services. Our primary aim in this project was the development of an approach for capturing a wide and comprehensive range of emotions from sparse, text based messages in social-media, such as Twitter, to help monitor emotional responses to events. Prior work has focused mostly on negative / positive sentiment classification tasks, and although numerous approaches employ highly elaborate and effective techniques with some success, the sentiment or emotion granularity is generally limiting and arguably not always most appropriate for real-world problems. In this paper we employ an ontology engineering approach to the problem of fine-grained emotion detection in sparse messages. Messages are also processed using a custom NLP pipeline, which is appropriate for the sparse and informal nature of text encountered on micro-blogs. Our approach detects a range of eight high-level emotions; anger, confusion, disgust, fear, happiness, sadness, shame and surprise. We report f-measures (recall and precision) and compare our approach to two related approaches from recent literature. © 2013 IADIS.

This is a conference paper. ; Social-Media streams are constantly supplying vast volumes of real-time User Generated Content through platforms such as Twitter, Facebook, and Instagram, which makes it a challenge to monitor and understand. Understanding social conversations has now become a major interest for businesses, PR and advertising agencies, as well as law enforcement and government bodies. Monitoring of social-media allows us to observe large numbers of spontaneous, real-time interactions and varied expression of opinion, often fleeting and private. However, human, expert monitoring is generally unfeasible due to the high volumes of data. This has been a major reason for recent research and development work looking at automated social-media monitoring systems. Such systems often keep the human "out of the loop" as an NLP (Natural Language Processing) pipeline and other data-mining algorithms deal with analysing and extracting features and meaning from the data. This is plagued by a variety of problems, mostly due to the heterogenic, inconsistent and context-poor nature of social-media data, where as a result the accuracy and efficacy of such systems suffers. Nevertheless, automated social-media monitoring systems provide for a scalable, streamlined and often efficient way of dealing with big-data streams. The integration of processing outputs from automated systems and feedback to human experts is a challenge and deserves to be addressed in research literature. This paper will establish the role of the human in the social-media monitoring loop, based on prior systems work in this area. The focus of our investigation will be on use of visualisations for effective feedback to human experts. A specific, custom built system's case-study in a social-media monitoring scenario will be considered and suggestions on how to bring back the human "into the loop" will be provided. Also some related ethical questions will be briefly considered. It is hoped that this work will inform and provide valuable insight to help improve development of automated social-media monitoring systems.

This is a conference paper. ; Due to the real-time nature and the value of social media content for monitoring entities and events of significance, automated sentiment analysis and semantic enrichment techniques for social media streams have received considerable attention in the literature. These techniques are central to monitoring social-media content, which is now becoming a significant business with commercial, institutional, governmental and law enforcement interest into its applications. Prior work in sentiment analysis especially has focused mostly on negative-positive sentiment classification tasks. Although numerous approaches employ highly elaborate and effective techniques with some success, the sentiment or emotion granularity is generally limiting and arguably not always most appropriate for real-world problems. In this paper a newly developed ontology based system is employed, to semantically enrich Tweets with fine-grained emotional states, in order to analyse the subjective public reactions to a wide selection of recent events. The approach detects a range of eight high-level emotions and their perceived strength (also known as activation level), specifically; anger, confusion, disgust, fear, happiness, sadness, shame and surprise. A set of emotional profiles for different events is obtained and an in-depth analysis of the emotional responses is presented. Recent events, such as the 2013 horsemeat scandal, Nelson Mandela's death, September 11th remembrance anniversary, recent tube strikes in London are analysed and discussed. The feasibility and potential benefits of automated fine-grained emotional event response analysis from social-media is illustrated and further, future work suggested.