“Come Together!”: Interactions of Language Networks and Multilingual Communities on Twitter

Abstract

Emerging tools and methodologies are providing insight into the factors that promote the propagation of information in online social networks following significant activities, such as high-profile international social or societal events. This paper presents an extensible approach for analysing how different language communities engage and interact on the social networking platform Twitter via an analysis of the Eurovision Song Contest held in Stockholm, Sweden, in May 2016. By utilising language information from user profiles (N =1,226,959) and status updates (N =7,926,746) to identify and categorise communities, our approach is able to categorise these interactions, as well as construct network graphs to provide further insight on these multilingual communities. The results show that multilingualism is positively correlated with activity whilst negatively correlated with posting in the user's own language.

Modelling and Analysing Behaviours and Emotions via Complex User Interactions

“…New and larger datasets: The recent events occurs regarding Facebook, suggested to explore other social networks (e.g. Twitter, LinkedIn) to extract a new dataset, number of companies are using Twitter as a main context of the communication between their customers and commonly used to report technical issues, Twitter offers a rich API [198] makes it a productive environment for researchers to collect and analyse large-scale longitudinal datasets [5,6,4]. Furthermore, since Twitter is largely an open, public platform the data can be used in investigating further and verifying the model outcome and improving the accuracy of the model.…”

“…These multiple interactions can be measured, profiled and modelled to critically analyse the communications and related processes and provide valuable insight into improving system architectures and designs. Furthermore, it provides a predictive capability to better understand digital behaviours -particularly through big social media datasets and corpora [186,271,220,45,5,6,4] -and thus develop systems that are more resilient and robust against undesirable behaviour and security breaches, such as "insider threat" scenarios [168,236,331], cyberhate [46,352], as well as more generally for crime informatics [238,47,237,239]. [143,266,112,159,51]) This research project is grounded in the area of human-computer interaction where it intercepts with the fields of artificial emotional intelligence and behavioural sciences under the umbrella of computer science (see Figure 1.1).…”

An Evaluation of Performance and Competition in Customer Services on Twitter

“…It provides a rich, constantly updating, corpus of big social data to study a range of complex socio-cultural issues, from life event detection [5] and identifying multilingual communities [2], through to sentiment classification [4] and providing deeper insight into personality and behaviour [17]. Unsurprisingly, Twitter is increasingly being used by organisations to communicate with their customers, due to the fast and convenient medium of engagement [15], using a variety of sophisticated human and automated approaches [25,27].…”

Popularity and Geospatial Spread of Trends on Twitter: A Middle Eastern Case Study

“…A common approach in analysing Twitter trends is through clustering and classification of trending topics based on content [13][14][15][16]. The study in [17] presented a content-independent method to model trends progression through the dynamics of users interactions; other studies have also attempted to provide real-time classification or detection of trends [18,19].…”