Matthew Rowe

Approaches to visualising linked data: a survey

Abstract. The uptake and consumption of Linked Data is currently restricted almost entirely to the Semantic Web community. While the utility of Linked Data to non-tech savvy web users is evident, the lack of technical knowledge and an understanding of the intricacies of the semantic technology stack limit such users in their ability to interpret and make use of the Web of Data. A key solution in overcoming this hurdle is to visualise Linked Data in a coherent and legible manner, allowing nondomain and non-technical audiences to obtain a good understanding of its structure, and therefore implicitly compose queries, identify links between resources and intuitively discover new pieces of information. In this paper we describe key requirements which the visualisation of Linked Data must fulfil in order to lower the technical barrier and make the Web of Data accessible for all. We provide an extensive survey of current efforts in the Semantic Web community with respect to our requirements, and identify the potential for visual support to lead to more effective, intuitive interaction of the end user with Linked Data. We conclude with the conclusions drawn from our survey and analysis, and present proposals for advancing current Linked Data visualisation efforts. Keywords: Linked Data, information visualisation, visual analytics, user-centred design, users, consumption

Modelling and analysis of user behaviour in online communities

Understanding and forecasting the health of an online community is of great value to its owners and managers who have vested interests in its longevity and success. Nevertheless, the association between community evolution and the behavioural patterns and trends of its members is not clearly understood, which hinders our ability of making accurate predictions of whether a community is flourishing or diminishing. In this paper we use statistical analysis, combined with a semantic model and rules for representing and computing behaviour in online communities. We apply this model on a number of forum communities from Boards.ie to categorise behaviour of community members over time, and report on how different behaviour compositions correlate with positive and negative community growth in these forums.

The effect of user features on churn in social networks

Social sites and services rely on the continuing activity, good will and behaviour of the contributors to remain viable. There has been little empirical study of the mechanisms by which social sites maintain a viable user base. Such studies would provide a scientific understanding of the patterns that lead to user churn (i.e. users leaving the community) and the community dynamics that are associated with reduction of community members -- primary threats to the sustainability of any service. In this paper, we explore the relation between a users value within a community - constituted from various user features - and the probability of a user churning.

Predicting discussions on the social semantic web

Social Web platforms are quickly becoming the natural place for people to engage in discussing current events, topics, and policies. Analysing such discussions is of high value to analysts who are interested in assessing up-to-the-minute public opinion, consensus, and trends. However, we have a limited understanding of how content and user features can influence the amount of response that posts (e.g., Twitter messages) receive, and how this can impact the growth of discussion threads. Understanding these dynamics can help users to issue better posts, and enable analysts to make timely predictions on which discussion threads will evolve into active ones and which are likely to wither too quickly. In this paper we present an approach for predicting discussions on the Social Web, by (a) identifying seed posts, then (b) making predictions on the level of discussion that such posts will generate. We explore the use of post-content and user features and their subsequent effects on predictions. Our experiments produced an optimum F1 score of 0.848 for identifying seed posts, and an average measure of 0.673 for Normalised Discounted Cumulative Gain when predicting discussion levels.

Who will follow whom? exploiting semantics for link prediction in attention-information networks

Existing approaches for link prediction, in the domain of network science, exploit a networks topology to predict future connections by assessing existing edges and connections, and inducing links given the presence of mutual nodes. Despite the rise in popularity of Attention-Information Networks (i.e. microblogging platforms) and the production of content within such platforms, no existing work has attempted to exploit the semantics of published content when predicting network links. In this paper we present an approach that fills this gap by a) predicting follower edges within a directed social network by exploiting concept graphs and thereby significantly outperforming a random baseline and models that rely solely on network topology information, and b) assessing the different behaviour that users exhibit when making followee-addition decisions. This latter contribution exposes latent factors within social networks and the existence of a clear need for topical affinity between users for a follow link to be created.

Ontology paper: Community analysis through semantic rules and role composition derivation

Online communities provide a useful environment for web users to communicate and interact with other users by sharing their thoughts, ideas and opinions, and for resolving problems and issues. Companies and organisations now host online communities in order to support their products and services. Given this investment such communities are required to remain healthy and flourish. The behaviour that users exhibit within online communities is associated with their actions and interactions with other community users while the role that a user assumes is the label associated with a given type of behaviour. The domination of one type of behaviour within an online community can impact upon its health, for example, it might be the case within a question-answering community that there is a large portion of expert users and very few users asking questions, thereby reducing the involvement of and the need for experts. Understanding how the role composition - i.e. the distribution of users assuming different roles - of a community affects its health informs community managers with the early indicators of possible reductions or increases in community activity and how the community is expected to change. In this paper we present an approach to analyse communities based on their role compositions. We present a behaviour ontology that captures user behaviour within a given context (i.e. time period and community) and a semantic-rule based methodology to infer the role that a user has within a community based on his/her exhibited behaviour. We describe a method to tune roles for a given community-platform through the use of statistical clustering and discretisation of continuous feature values. We demonstrate the utility of our approach through role composition analyses of the SAP Community Network by: (a) gauging the differences between communities, (b) predicting community activity increase/decrease, and (c) performing regression analysis of the post count within each community. Our findings indicate that communities on the SAP Community Network differ in terms of their average role percentages and experts, while being similar to one another in terms of the dominant role in each community - being a novice user. The findings also indicate that an increase in expert users who ask questions and initiate discussions was associated with increased community activity and that for 23 of the 25 communities analysed we were able to accurately detect a decrease in community activity using the communitys role composition.

Linked knowledge sources for topic classification of microposts: A semantic graph-based approach

Short text messages, a.k.a microposts (e.g., tweets), have proven to be an effective channel for revealing information about trends and events, ranging from those related to disaster (e.g., Hurricane Sandy) to those related to violence (e.g., Egyptian revolution). Being informed about such events as they occur could be extremely important to authorities and emergency professionals by allowing such parties to immediately respond. In this work we study the problem of topic classification (TC) of microposts, which aims to automatically classify short messages based on the subject(s) discussed in them. The accurate TC of microposts however is a challenging task since the limited number of tokens in a post often implies a lack of sufficient contextual information. In order to provide contextual information to microposts, we present and evaluate several graph structures surrounding concepts present in linked knowledge sources (KSs). Traditional TC techniques enrich the content of microposts with features extracted only from the microposts content. In contrast our approach relies on the generation of different weighted semantic meta-graphs extracted from linked KSs. We introduce a new semantic graph, called category meta-graph. This novel meta-graph provides a more fine grained categorisation of concepts providing a set of novel semantic features. Our findings show that such category meta-graph features effectively improve the performance of a topic classifier of microposts. Furthermore our goal is also to understand which semantic feature contributes to the performance of a topic classifier. For this reason we propose an approach for automatic estimation of accuracy loss of a topic classifier on new, unseen microposts. We introduce and evaluate novel topic similarity measures, which capture the similarity between the KS documents and microposts at a conceptual level, considering the enriched representation of these documents. Extensive evaluation in the context of Emergency Response (ER) and Violence Detection (VD) revealed that our approach outperforms previous approaches using single KS without linked data and Twitter data only up to 31.4% in terms of F1 measure. Our main findings indicate that the new category graph contains useful information for TC and achieves comparable results to previously used semantic graphs. Furthermore our results also indicate that the accuracy of a topic classifier can be accurately predicted using the enhanced text representation, outperforming previous approaches considering content-based similarity measures.

The OU linked open data: production and consumption

The aim of this paper is to introduce the current efforts toward the release and exploitation of The Open Universitys (OU) Linked Open Data (LOD). We introduce the work that has been done within the LUCERO project in order to select, extract and structure subsets of information contained within the OU data sources and migrate and expose this information as part of the LOD cloud. To show the potential of such exposure we also introduce three different prototypes that exploit this new educational resource: (1) the OU expert search system, a tool focused on finding the best experts for a certain topic within the OU staff; (2) the Social Study system, a tool that relies on Facebook information to identify common interest between a users profile and recommend potential courses within the OU; and (3) Linked OpenLearn, an application that enables exploring linked courses, Podcasts and tags to OpenLearn units. Its aim is to enhance the browsing experience for students, by detecting relevant educational resources on the fly while studying an OpenLearn unit.

The credibility of digital identity information on the social web: a user study

The recent rise in the adoption of Social Web platforms such as MySpace, Facebook and Twitter has provided Web users with rich functionality and feature sets to interact with their peers and construct an online presence. The digital identity which Web users build on the Social Web is being increasingly reused by third party services (product recommendation services, authentication mechanisms, identity management services). The reliance on such digital identity information requires accurate and credible information. This paper presents a detailed user study of the digital identity representations which are constructed on the Social Web. The study explores the extent to which such representations mirror their real-world equivalent and therefore assesses the credibility of such information.

Mining and comparing engagement dynamics across multiple social media platforms

Understanding what attracts users to engage with social media content (i.e. reply-to, share, favourite) is important in domains such as market analytics, advertising, and community management. To date, many pieces of work have examined engagement dynamics in isolated platforms with little consideration or assessment of how these dynamics might vary between disparate social media systems. Additionally, such explorations have often used different features and notions of engagement, thus rendering the cross-platform comparison of engagement dynamics limited. In this paper we define a common framework of engagement analysis and examine and compare engagement dynamics, using replying as our chosen engagement modality, across five social media platforms: Facebook, Twitter, Boards.ie, Stack Overflow and the SAP Community Network. We define a variety of common features (social and content) to capture the dynamics that correlate with engagement in multiple social media platforms, and present an evaluation pipeline intended to enable cross-platform comparison.Our comparison results demonstrate the varying factors at play in different platforms, while also exposing several similarities.