Emerald Group Publishing Limited
Copyright © 2011, Emerald Group Publishing Limited
Katrin Weller's “Knowledge Representation in the Social Semantic Web” is intended to “help close the gap between knowledge representation methods in the Social and the Semantic Web” (p. 9). The gap is between the uncontrolled tags of folksonomies and highly formalised ontologies of Semantic Web research. This is done through an initial review of classical knowledge organisation principles and practice, which is followed by an overview of ontologies, leading to the final chapter which considers KOS and ontologies in relation to the social web with a view to bridging folksonomies and ontologies. The hope is that the book will act as “another little step in the long path towards the ultimate goal” (p. 9), which is closing the semantic gap and dealing with information overload. It's a big broad aim for a big, heavyweight book in which Weller provides information, examples, extensive bibliographies, as well as some great starting points for possible research projects. The book makes some assumptions about the readership, which is deemed to have an already existing understanding of traditional KOS, and at least a vague understanding of what ontologies are and how they work. Some important topics are skirted over, assumed already known. On the other hand, I found some of the fairly detailed discussion of ontology methodology in the second chapter a little heavy‐goiug.
The book is organised into three main – long – chapters. The first (pp. 18‐114) considers classical KOS, the second (pp. 115‐240) focuses on ontologies, while the third (pp. 241‐402) is entitled “Ontology engineering in the Era of the Social Semantic Web” and in this chapter, the principles of ontology engineering are placed in the context of the Social Web. I found this to be the most interesting chapter in the book as it contains, amongst the near‐overload of information, some very interesting ideas for future research projects. Each chapter includes an extensive list of references, for example, the references for chapter 1 run from page 95 to page 114.
The first chapter fairly gallops through a discussion of traditional KOS, covering the essentials of classification, faceted classification, and thesauri, in about nine pages, including examples. The discussion then moves on to indexing the web, and by page 53, we reach “The Semantic Web”. The final section of chapter 1 considers the notion of the social web and the concept of Web 2.0. Here the discussion explores the convergence of the semantic web and Web 2.0 that already exists in the form of semantic wikis and blogs. Weller, referring back to Gruber (2005), emphasises the main point of her thesis, which is that ontologies and folksonomies should not be seen as rivalling systems but as potentially complementary systems, enriching approaches to knowledge organisation on the web.
The second chapter starts by relating the concept “ontology” in AI and computer and information sciences to its origins in philosophy, and the point is made that the deliberations involved in developing the upper‐level ontologies that deal with abstract essences making up World Knowledge are still very close and make reference to philosophical traditions (p. 116). The difficulties involved in defining precisely this something slippery concept are acknowledged. For information science, discussion about the precise nature and function of ontologies sometimes focuses on the relationship between ontologies and other types of KOS (a point that Weller returns to in some detail later in the book), while for computer science, the discussion is often about the relationship between ontologies and databases (p. 119). Weller writes that ontologies should “Unambiguously represent shared background knowledge that helps people within a community of interest to understand each other. And they should make computer‐readable indexing of information possible on the Web. All this has not been the case for traditional KOS and it is an important innovation” (p. 123). One might ask for more discussion here. Could not faceted classifications form such a base, drawing from classical KOS? Are ontologies really unambiguous? What might be the limits of consensus and inter‐subjective meaning and understanding within an “community of interest”, and what precisely constitutes such an entity? Quite often, while I was reading this book, I felt that there were assumptions being made about the meaning, understanding, and communication that tend towards the functional, seeing communication as relatively unproblematic. It might be that Weller, who has a background in linguistics, so is not without knowledge of the complexities of communication, has decided that adopting this position is useful to move the central argument of the book along‐ it would certainly be easy to become waylaid by arguments about meaning and interpretation, but there were times when I wanted her to slow down and explore the broader issues a little more.
Weller, following Stock (2007), refers to the semiotic triangle of Ogden and Richards (1969) in her exploration of ontological concepts and classes. It is always good to see semiotics used explicitly as a theoretical foundation for Knowledge Organisation, but the omission of the broader history of semiotic traditions is noticeable. The semiotic concepts paradigm and syntagm, as employed by Stock, are used to discuss some of the semantic relations in KOS, but the way that Stock seems to use these terms is fairly rigid and specific, and not necessarily the way that they are used in other versions of semiotics, or indeed in other versions of human information retrieval (e.g. Warner, 2010). In the discussion that follows, which moves from a consideration of top‐level ontologies, to a discussion of the difficulties of determining concepts, their attributes and concept hierarchies, the Bliss Classification working group's 13 fundamental categories is referred to (p. 142), reminding us that many of the theoretical issues have long been considered in the context of classical KOS.
Much of this chapter contains fairly technical and detailed discussion of the structure, functions, and creation of ontologies. Different types of semantic relationships are considered – those of equivelance, hierarchy, association, metonymy, hyponymy – and the point is made that unlike the more standardised associative relationships we find in traditional KOS, ontology engineers have the freedom to extend arbitrarily (p. 146), providing rich representations. I would imagine that the audience for this part of the book is considered to be an information science readership which, while it is strong on KOS and related domains, is perhaps less expert on ontology engineering. While Weller explains the various relationships fairly clearly, the English is sometimes a little stilted and lacks smoothness, which can make the reading hard going at times. Towards the end of the chapter, Weller produces a typology of various types of KOS and distinguishes between them in terms of their structures. While this is all very interesting, and is certainly a useful analysis, the structure of this section of the chapter is a little bit plodding, a legacy perhaps of its genesis as an article. It also feels a little repetitious in places. However, the analysis leads to the formulation of some interesting questions relating to whether it is possible to harness the wisdom of crowds to do ontology engineering work, and these questions form the focus of the third and most interesting chapter, which considers how to expand ontology engineering so that ontologies can be built by groups of users in collaboration.
The chapter begins by considering the steps to be undertaken in ontology engineering methodology and by describing some of the functionalities of ontology editors. The point is made that in future work it would be sensible to reuse existing ontologies, and to move towards automatic ontology engineering. Ontology management tools focus on the development of ontologies rather than on ontologies for indexing, and questions of indexing poses new challenges to ontology tools (p. 267). One of the “all‐time hot topics” (p. 271) for ontology engineering is the “support of user collaboration” in the form of community‐based ontology engineering. Such an undertaking would, ideally, combine the expertise of the ontology engineer with the knowledge expertise of the community of experts. Ways in which such participation might be managed are considered, with examples of some collaborative ontology engineering tools used to illustrate possibilities. Games with a purpose are discussed, and examples such as Google Image labeller are included (p. 306). Following this really interesting section, Weller then considers the issues of reusing existing KOS data to create ontologies, and she does this through the extended metaphor of tag gardening, so the activities to be considered in the process of enhancing and enriching folksonomies by adding semantic structure are labelled: seeding, weeding, landscape architecture and fertilizing. The extended metaphor, based on previous work, is perhaps a little irritating after a while, but there is some extremely useful information in this section.
The argument moves on to considering ways of integrating multiple KOS on the web to enable the broadest possible coverage. Can we “mediate between divergent points of view” (p. 351) Weller asks; can we manage the many KOS that come from different types of user groups and different sized user groups, that might even represent one person's perspective? Increasingly, the focus is on “inter‐ontology engineering” (p. 351), and the different KOS interactions will include reuse, upgrade, matching and mapping. KOS interoperability might come through interaction with upper ontologies, or different KOS might be collected together to form a new and broader KOS – a signal for Weller to discuss her own KOSO (Knowledge Organisation Systems Ontology) meta‐KOS project (p. 361). A brief mention of SKOS (Simple Knowledge Organization System) and the Linked Data initiative follows. I think in this context, these projects really merit a little more discussion that they are given here. As with some of the chapter 1 material, some topics are dealt with in a relatively sketchy way. But overall, this is a great chapter. It is rich with information, but beyond that, it provides ideas for possible research topics. Sometimes the tone seems to me to be overly certain about the possibility of consensus and agreement: there are, however, promising gaps, chinks in the certainty, for example the need to acknowledge divergent opinions, the notion of personomy, which could provide openings for critical research work.
In summary, this is a heavyweight book. It is very scholarly and is rich with information and with ideas. It provides the reader with an excellent bibliography and it contains a useful index. The English is sometimes a little clumsy, and there is a tendency towards repetition, but, nevertheless, in the end the book is worth the effort.
Gruber, T. (2005), “Collective knowledge systems: where the Social Web meets the Semantic Web”, Journal of Web Semantics, Vol. 6 No. 1, pp. 4‐13.
Ogden, C.K. and Richards, I.A. (1969), The Meaning of Meaning: A Study of the Influence of Language upon Thought and of the Science of Symbolism, Routledge and Kegan Paul, London (originally published in 1923).
Stock, W.G. (2007), Information Retrieval: Informationen Suchen und Finden, Oldenbourg, München/Wien.
Warner, J. (2010), Human Information Retrieval, The MIT Press, Cambridge, MA.