Undefined 0 (0) 1 1 IOS Press A Review of Argumentation for the Social Semantic Web Jodi Schneider a, Tudor Groza a;b, and connected arguments posted by individuals to express Alexandre Passant a their opinions in a structured manner” [5]. a Digital Enterprise Research Institute, National Arguments on the Web can be used in decision con- University of Ireland, Galway, texts. Decision-making often requires discussion not fi[email protected] just of agreement and disagreement, but also the prin- b School of ITEE, The University of Queensland, ciples, reasons, and explanations driving the choices Australia, [email protected] between particular options. Furthermore, arguments expressed online for one audience may be of inter- est to other (sometimes far-flung) audiences. And it can be difficult to re-find the crucial turning points of an argumentative discussion in which we have partic- Abstract. Argumentation represents the study of views and ipated. Yet on the Web, we cannot subscribe to argu- opinions expressed by humans with the goal of reaching a ments or issues, nor can we search for them. Nor can conclusion through logical reasoning. Beginning with the we summarize the rationale behind a group’s decision, 1950’s, several models were proposed to capture the essence even when the discussion took place entirely in public of informal argumentation in different settings. With the venues such as mailing lists, blogs, IRC channels, and emergence of the Web, and then the Semantic Web, this Web forums. modeling shifted towards ontologies, while from the de- velopment perspective, we witnessed an important increase By providing common languages and principles to in Web 2.0 human-centered collaborative deliberation tools. model and query information on the Web (such as Through a review of more than 150 scholarly papers, this ar- RDF [6], RDFS [7], OWL [8], SPARQL [9], Linked ticle provides a comprehensive and comparative overview of Data principles [10], etc.), the Semantic Web [11] is the argumentation domain for the Social Semantic Web. We an appropriate means to represent arguments and ar- start from theoretical foundational models and investigate gumentation uniformly on the Web, and to enable, for how they have influenced Social Web tools. We also look into instance, browsing distributed argumentation patterns Semantic Web argumentation models. Finally we end with that appear in various places on the Web. Indeed, re- Social Web tools for argumentation, including online appli- searchers have shown that the Semantic Web can be cations combining Web 2.0 and Semantic Web technologies, used for visualization and comparison in decision ra- following the path to a global World Wide Argument Web. tionale [12]. Keywords: Argumentation, Semantic Web, Social Web, Se- In this context, this paper discusses argumentation mantic Web, Ontologies in relation to the Social Semantic Web [13,14,15], fo- cusing on foundational models of argumentation, their applications in the Social Web, and on ontologies (as 1. Introduction in Computer Science [16]). In particular, our purpose is to investigate ontolo- In recent years, large-scale argumentation on the gies and tools which may be useful for argumentation Web has attracted the attention of scholars from fields on the Social Semantic Web, a field where the afore- such as artificial intelligence [1], communication the- mentioned Semantic Web technologies support Social ory [2], business management [3] and e-government Web [17] applications, while at the same time Social [4]. At the same time, argumentation researchers be- Web paradigms are used to generate Semantic Web gan establishing the foundations for a World Wide Ar- data collaboratively and at large scale. This conver- gument Web (WWAW) as “a large-scale Web of inter- gence aims at providing new and improved ways to 0000-0000/0-1900/$00.00 c 0 – IOS Press and the authors. All rights reserved 2 Schneider et al. / A Review of Argumentation for the Social Semantic Web integrate and discover data, following the vision of we convince ourselves and others of our points of Social Machines provided by Berners-Lee [18], both view. Informal argumentation occurs throughout con- on the Web and in the enterprise [19]. In the context versations, online and off, often in conjunction with of argumentation, this could help to aggregate argu- persuasion or with joint decision-making. Social me- ments from various websites — for instance a discus- dia’s diverse decision-making happens in many online sion starting on Twitter and followed up on a mailing discussion fora such as standardization bodies’ list- list, later frozen on a wiki once consensus is reached servs, Wikipedia editors’ wiki pages, and open source — providing new means to follow argumentative dis- communities’ IRC channels, bug reports, and listservs. cussions on the Web. This would enable an argument- Even logically sound decisions may involve choices centric view of the Web. based on values and preference judgements: people Moreover, the Social Web does not yet have widely- may agree on the facts of a situation yet disagree on used argumentative ontologies, though this problem the preferred outcome or decision to be taken. That is has been noted [20], along with the need for federation vitally different from disagreeing on the facts of a sit- infrastructures [21]. Thus, in order to identify how dif- uation (in which case more information is called for). ferent argumentation models and tools can be used for the Social Semantic Web, this paper offers a review of more than 150 research papers on the topic, from 1945 to 2011, from which we compare: – 13 theoretical models of argumentation – 14 Semantic Web models for argumentation (i.e. ontologies) – 37 tools for representing argumentation on the Web. Fig. 1. Common argument patterns, from [23]. As the focus is on human-centered argumentation [22], with the goal of improving access and visualiza- There are a variety of common argument structures tion, this article will briefly mention, but not analyze, [23]. A single premise may directly support a conclu- the agent-based argumentation domain. sion (as in Figure 1(i)), but more commonly, they are Following the introduction, we provide brief overviews combined to come to a conclusion. Premises and con- of argumentation (Section 2.1) and of the Social Web clusions may also be chained (as in Figure 1(iv)). (Section 2.2), then discuss requirements for support- ing argumentation on the Social Semantic Web (Sec- 2.2. Social Semantic Web tion 3). We next present theoretical models of argu- mentation (Section 4) from a variety of fields, compare The interaction of users around the Web has been them (Section 5), and present applications of these the- shifting from individual siloed Web systems, towards oretical models (Section 6). Subsequently we present more open and interlinked social applications1. In dis- (Section 7) and compare (Section 8) Semantic Web cussion environments, such interlinkage is particu- models of argumentation. Then we move on to review- larly important: the same community may discuss top- ing tools: in Section 9 we highlight thirteen notewor- ics across multiple sites, and use multiple types of thy features of Social Web argumentation tools, based sites, such as blogs and microblogs, discussion forums, on a comprehensive analysis of thirty-seven relevant and wikis. Crosslinking the discussions of these sys- tools (see the Appendix for full details). Finally we tems is a first step, which has been taken by SIOC conclude the paper in Section 10. – Semantically-Interlinked Online Communities [24]. Yet the internal structure of these discussions – such 2. Background as whether the participants agree or disagree, are con- tributing diverse ideas, or debating in circles – is still 2.1. Argumentation not represented in SIOC. Capturing such underlying Argumentation is the study of agreement, disagree- 1http://oreilly.com/web2/archive/ ment, and of the dialogues and writing through which what-is-web-20.html Schneider et al. / A Review of Argumentation for the Social Semantic Web 3 arguments would be valuable, and research is begin- ception, refinement, and pronomial anaphora and call ning to address this for instance by identifying argu- for a modular architecture “where different relation- ment schemes used in Amazon reviews [25] and by ships or debate components may be added systemati- modeling the speech acts in Twitter conversations [26]. cally” [27]. Yet infrastructure for argumentation on the Social Se- mantic Web is still needed. 3.1. Example Applications & Requirements We envision two main approaches to argumentation 3. Requirements on the Social Semantic Web: 1. Focusing on the real-time, dialogical nature of What are the requirements for supporting argu- the Social Web, i.e. by soliciting arguments from mentation on the Social Semantic Web? Arguments humans through conversation and real-time ex- must be identified, resolved, represented and stored, change. queried, and presented to users. Identification involves 2. Focusing on the Social Web as a source of ar- mining arguments, in the form of claims, from text tifacts, i.e. by using existing natural language (Section 6.11.2, page 15), eliciting them from users, conversations and reconfiguring the traces and or some combination of these approaches. Resolving archives of these conversations. involves indicating the relationships between the indi- vidual claims that make up arguments: are they on the Examples of the first case would be a chatbot or same topic? Do they agree or disagree? Representing an interactive webform; these could help populate a and Storing arguments uses a suitable ontology to rep- knowledge base or enable argumentative interaction resent claims and the relationships between them. This between humans and intelligent agents. Examples of supports Querying and enables Presenting the Social the second case would be summaries or interactive Semantic Argument Web, i.e. using these ontologies conversation browsers; a discussion summary could to facilitate access to conversations, summarizing the highlight the agreement and disagreement about a contentious and agreed-upon points of a discussion.