Folksonomological Reification Patrick Parslow, Karsten Ă˜ster Lundqvist, Edwin Porter-Daniels, Richard Hussey, Shirley Williams University of Reading
Abstract The gap between formal ontologies and folksonomies is identified, and termed the “Ontology-Folksonomy Divide�. This is explored through the application of sociological and psychological models of identity, which are examined and applied to the phenomena of online social networking, using a specially developed analysis framework based on activity theory. Descriptions of roles present in online communities and how they relate to an individual’s digital identity are explored and a taxonomy of roles is proposed. The methods, technologies, and theory lead to the idea of Folksonomological Reification.
Introduction: Why Folksonomological Reification? “The positive development of a society in the absence of creative, independently thinking, critical individuals is as inconceivable as the development of an individual in the absence of the stimulus of the community” – Albert Einstein. There exists an inherent divide between the Web2.0 community and the knowledge representation / Semantic Web community. There are a large number of Internet services and tools, collectively termed Web2.0, that empower individuals to use their creative powers which, as a consequence, builds up a community knowledge base. This form of communal knowledge base, created by the individual actions of many people is termed a folksonomy. There is also a formalised approach to knowledge representation, called ontology, which can be utilised by computers in intelligent comparisons and inferences. This latter option is problematic due to the difficulty of developing ontologies. This is particularly true in scenarios where many people are involved, and thus this method is being under-utilised. It would be both interesting and beneficial if these two approaches could be connected, so that the semantics that are created in folksonomies could be utilised in computational semantic tools. This kind of technology does not yet exist, and this is why we have coined the “Ontology-Folksonomy Divide” (Figure 1) to explain this lack of interoperability. This chapter explores the issues with pursuing this line of enquiry, and investigate the roles and identities which individuals can take when doing so.
Figure 1 Ontology-Folksonomy Divide
Ontologies and FolksonomiesOntology “Ontology” has multiple definitions relating to different domains, although they are related. It is the study of being, but it is also used in a more technical regard as being the study of things and their relationships, and it is also used as a noun to mean a formal specification of a domain (Figure 2). As a noun it defines and specifies the different classes of individuals that form the domain, the actual individuals and the properties (relationships) of the individuals (Horridge, et al., 2004)
Figure 2 Ontology
One of the benefits of developing an ontology is that multiple applications can consistently use the same domain to perform different automated tasks, including tasks that involve reasoning based on the different relationships that are specified. Ontologies are rigid specifications, i.e. relationships either exist or do not exist, and there is no space for fuzziness or probabilities. The binary nature of ontologies is more technically described by Horrocks as the “crispness of ontologies” (Stoilos, et al., 2005), and can lead to issues when attempting to utilise their properties on a larger scale, as the users need to agree upon common ontologies. This has proved to be difficult, as most domains have experts and users disagreeing about some of the properties inherent in the domain. Fuzzy ontologies (Figure 3) have been proposed in the past (Lee, et al., 2005) (Parry, 2004) as a method to reduce the crispness of ontologies, and though they introduce degrees of trust in the different component parts of the knowledge base they are still only representing the knowledge from one viewpoint.
Figure 3 Fuzzy Ontology
Web2.0 Ontologies The nature of Web2.0 is fragmented and rapidly changing with an immense number of services and tools. Many of these tools internally utilise ontologies despite the fact that there are no optimal tools to semantically unite the disparate areas. From this viewpoint the term “Web 2.0 Ontologies” is not adequately defined, and thus does not lend itself well to the normal definition of the traditional ontology. It is therefore helpful to examine the nature of folksonomies, which are very much a part of the Web2.0 phenomenon and which provide access to some of the semantic features of ontologies whilst also inherently providing an interpretation of the fuzziness involved when many people share their own experiences. Thus, it should be possible to explore what is necessary to bridge the “Ontology-Folksonomy Divide” – the division between folksonomies and ontologies.
Folksonomies Folksonomies are descriptions or representations of domains, with emphasis on relationships made by individuals in the domain, gathered though processes performed by many people (Vander Wal, 2005) (Smith, 2008), such as interacting with ‘social software’. Therefore, folksonomies offer a certain ranking or fuzziness of how, for example, a “tag” is related to a web page. Unfortunately at present this folksonomy representation is only available to humans, e.g. as a view in a “tag cloud” or by manually searching through the knowledge base of the folksonomy to find interesting relationships.
“Social Software” The term “social software” can be used to cover a wide range of software tools that enable interaction between people (Boyd, 2003) and we describe two forms of socially-enabled software, which, whilst far from being mutually exclusive, illustrate two types of service, which support the emergence of folksonomies, i.e. social networking and social bookmarking. Marlow (Marlow, et al., 2006) gives a more detailed taxonomy of social tagging systems. Social networking sites are web-based tools facilitating social interaction. Whilst the facilities they provide vary, common core components are a profile page, the concept of ‘friends’ to whom the user has some form of online (and in some cases offline) links, and some method of making comments on material, whether it is internal or external to the site. Comments can have varying levels of privacy, taking the form of private communication between two individuals, one to many communications and many to many. Social bookmarking sites allow the users to post links to material they have ‘discovered’ on the web, and label the post with a combination of single word descriptions, or short phrases, known as tags. Some bookmarking sites allow for a comment as well as a set of tags (e.g. Clipmarks http://clipmarks.com/ and flickr http://flickr.com/). Whilst the tag is inherently less expressive than the comment available on social networking sites, the simplicity of the relationship between a single short descriptor and the resource with which it is associated provides a rich source of information when taken in conjunction with the tags provided by the other members of the community, and the tagging behaviours of those individuals.
The comment is amenable to similar treatment. Comments are typically written in something akin to normal language (differing through use of various online dialects) and interesting words can be automatically extracted by ignoring the ‘filler words’. Filler words currently have to be pre-defined, typically including conjunctives and pronouns, but once this is done the remaining set of words can be used as though they were tags. Whilst this undoubtedly loses some of the original meaning of the comment (by reducing it to a set of unordered key words), many of the semantic relationships between words and resource are maintained. The knowledge accumulated by several folksonomies based on the activities of single individuals could be represented using an ontology. By combining these ontologies, algorithms could be designed to navigate the total knowledge space of all the users (Figure 4). However, when using this approach, representation of the fuzziness of the knowledge on the individual level would still not be possible. Some shared knowledge could potentially remain as ‘separated’ knowledge due to the fact that disparate ontologies might use incompatible phrases and identifiers for it. This kind of system would also be prone to retaining redundant information.
Figure 4 Folksonomy represented by many ontologies
Alternatively, the body of knowledge could be stored in a crisp ontology that would describe everything, even diverging knowledge, thus making it searchable and consistent across different views of the domain. The knowledge would then be connected to different individuals by reifications in a folksonomic layer (Figure 5). This approach would allow diversity and make it possible to study emergent behaviours in the body of knowledge.
Figure 5 Folksonomy represented by one ontology using a reified folksonomological layer
Folksonomology We define folksonomology as “The study of emergent ontological properties arising from collaborative knowledge bases� and apply it to systems such as communal tagging services. The approach is to use the emergent properties from the implicit relationships existing in the tag clouds to create formalised semantics. These relationships loosely define the contextual environments of items that are shared within online communities. Through this context, and both the explicit and implicit relationships, we can establish meaning. Folksonomology is not restricted to analysing the emergent meaning of the resources that are tagged. In this chapter, it is argued that studying the tagging behaviours can give insight into the roles occupied by an entity, either an individual or a group, within a tagging enabled social network and give a view on the digital identities involved.
Folksonomological Reification Producing tools that help an individual to study these relationships, discuss them with their communities, and share their conclusions will enable Web2.0 users to undertake collaborative folksonomology. Hence we seek to reify the study of emergent ontological properties, or in other words, to perform folksonomological reification.
Theoretical Considerations Multiple strands of theory This chapter is based on several different areas of theory from different fields of academic endeavour. Folksonomies are an emergent semiotic phenomenon arising from communal activities stemming from the behaviour of the members of communities. Their study therefore requires elements from all these fields and provides fertile ground for exploring theory in them. This discussion relates to modern technologies, but can also be seen to have a bearing on more established fields of study. The folksonomy, for instance, may represent the development of new language – whilst it will typically re-use existing words and word-forms, the way they are used together, and the
meanings associated with them can be a reflection of the community who produce it – “The world won’t stay miscellaneous because we are together making it ours” (Weinberger, 2007). Studying the way individuals work independently, for their own reasons and to their own ends, but collaboratively by dint of the tools they choose to use, may give further insight into social identity theory and psychology.
Activity Theory In traditional Activity Theory, as expanded by Engeström (Engeström, 1999), the activity system is viewed as comprising six parts, which are often represented by the activity triangle (Figure 6):
Figure 6 The traditional Activity Theory triangular relationship mode
Whilst exploring different methods and theories to use in helping analyse ‘social software’, it was felt that the underlying dynamic nature of Activity Theory well suited the interactions between people, community, and tools found in the domain. A modified view of the Activity Theory relationship brings out the key relationships in the interactions between community and tools. Taking the multiple identity/multiple role aspects of the individual/community outlined above, and considering the Rules and Division of Labour of activity theory as defining the (formal and semi-formal) group Roles, the diagram can be simplified (Figure 7) for use in analysing the types of social communities, behaviours and tools which are the subject of study. The possible multiplicity of individual goals which make up the Object are grouped with the communal objectives. In this case, recognising the folksonomy as a dynamic tool, created and used by the members of the community, we place Tools at the centre of the diagram. In terms of the communities that form around social networks and social utilities, there would be no community without the tools, which is why there is no link in the model directly joining the individual Subjects with the Community. It is recognised that the framework may be useful in other scenarios, however, such as to help a Technology Steward analyse a potential technology for use by a community of practice. In these cases, the Object and the Community become more central, and a key concern is how to select the appropriate tools to provide a close linkage between the individual’s goals and the community’s objectives.
Figure 7 Revised Relationship model for Social Networking tools and Activities
Individuals and Communities The nature of the wisdom of the crowd Folksonomology is the study of the formation of rules by the ‘wisdom of the crowd’ (Surowiecki, 2004). This can enable the practitioner to gain insights into the meanings the folk (individuals acting separately) attach to things, and also allow for a view to be gained of the way they reach their agreements, and of the roles they play within the group. As the Web2.0 technologies become more endemic, and ease of many-to-many communication increases, the user will tend to become more of a ‘prosumer’, producing and consuming (web) content. Users of social bookmarking, mashup and social network sites already fall into this category, creating content for their own purposes which can then be consumed, repurposed and re-published by others. One should not underestimate the wisdom of the crowd effect nor regard it as being a newly recognised phenomenon. Galton (Galton, 1907) demonstrated that whilst the median value from a set of estimates by many people of the weight of a bull was within 1% of the true value, the distribution of estimates did not fit a normal distribution. In his analysis, he suggests that the observers were minimising some types of error, and maximising others. Of course, because the data from which he worked was produced as part of a 'guess the weight' competition, the estimates were likely to be independent, and that is not a feature of the type of data produced by communal bookmarking or other social networking enterprises. Here, the individuals who make decisions on which word to use as a tag for a piece of content have access to the existing corpus, although they may not all make use of it and consequently there would not be an expectation of a normal distribution.
Roles in communities supporting the wisdom of the crowd effect A taxonomy for discussion about roles in community is needed. Some of the proposed names for roles, with brief descriptions, serve to motivate the later discussion. Users who bookmark content for their own use will tend to use tags that mean something to themselves (Bookmarker). This behaviour is probably the most similar to the 'guess the weight' competition, and relies on the individual acting in their own self-interest. However, users sometimes tag in order to heighten awareness (Explorer) of content for other people. Sometimes this is in order to raise the profile of the material they have posted, and sometimes it is to bring the attention of a group to something which they think might, or should, be of interest to them. In these cases, the tagger can make use of tags which are pre-agreed with their community, or which may be found easily by others. These two cases are likely to generate either very specific tags (often comprising of amalgams of words and numbers), combinations of tags, or very generic ones. A third role the posting individual may take is that of an Editor - deliberately adding tags to content which seem to be appropriate but which are missing from the existing corpus. Fourthly, a Translator may add tags in languages other than the original, additionally there may be some disruptive individuals who 'poison' the system by adding inappropriate tags. These behaviour patterns are not necessarily easy to detect, but may have a significant impact on the collections of tags which are applied to content. The ability to see the existing set of tags for an item may also tend to make most users select from existing tags, especially where the system “helpfully” suggests commonly used ones. Whilst this may contribute to a more coherent body of tag-content pairings it may, in fact, have a detrimental effect on the value of the tagging exercise. The individual may accept a tag suggestion despite it not being the word they might have used if left to their own devices. The community may end up with a narrower focus in the tag cloud, which may not express the richness of the content adequately. The strongest 'defence' against the suggested tag is to foster the Editor role in users - people who will examine the existing tag set and expand it to include less obvious but still relevant tags.
People, Roles and Digital Identity There are different roles people can undertake when they engage in social bookmarking. In 'full-blown' social networking there are even more roles varying in terms of formality and complexity. These roles are not yet well defined, and the reader may well be able to identify many others or hold the nomenclature in dispute. The names, and indeed descriptions, are there to stimulate discussion - the 'final definitions' will come about as a folksonomy.
Identity is meaning; Multi-faceted individuals and social identity Taking the Identity Theory (Stryker, et al., 2000) view that identity is a reference to a part “of a self, composed of the meanings that persons attach to the multiple roles they typically play in highly differentiated contemporary societies” (Stryker, et al., 2000) we see that the roles we adopt form a large part of our concept of self, and that, as in the common parlance, they give meaning to our place in the group. These roles are reflections of elements of self which can be thought of as strands of identity. These strands can all co-exist within the individual, although only one is consciously active at
a time. Roles relating to family life, work, being a customer or a seller form the social identity of the individual. Those roles that are applicable to the individual’s digital environment form a digital social identity for that individual. This is the identity of the individual as viewed from the perspective of others in the digital environment, rather than the individual’s own view of their identity. But in many ways, it is these views which go to forming the individual's identity, as it is these views which will determine the social interactions of the individual. Merchant gives a good analysis of issues relating to identity and online social networking (Merchant, 2006). Through our interactions with others in social networks, whether online or offline, we create meaning by classifying and labelling resources. These resources can be of any type, including people. Whilst doing so we occupy different roles, presenting different strands of ourselves, different identities, to the community in which we are involved. The meaning of these identities to other members of the group goes towards establishing their views of our identity, and the meaning we attach to the identities we present forms our view of our self. But these meanings are themselves informed by the views of our identity expressed by other members of the group, whether explicitly or implicitly. This is strongly reminiscent of Symbolic Interactionism (Blumer, 2004), wherein the self is seen as a construct of society. Clarke describes the digital persona (Clarke, 1993) as the model of an individual which can be produced by the collection and analysis of information about that person. This digital persona is usually considered in terms of the models, which can be built by computer programs, but the concept holds just as well for the model built by human agents who have access to the digital information. This digital persona, then, is the overall model which comprises the sum of the digital identities, each representing a strand of the biological individual’s personality in terms of the roles they perform in different contexts.
The individual as part of the community Any community consists of norms (which include hierarchical rules of behaviour and cultural prejudices, amongst many others), a lingo, ideals and shared experience (Van Maanen, et al., 1979) and the social network, and social networking communities (Shirky, 2008) are not significantly different (Benkler, 2006). The lingo is an emergent property, and can be seen to evolve within different sub-cultures of the whole, both in online and offline communities. The social norms vary, with some ground rules for behaviour being laid down by the owners or founders of the community. Some of these are more restrictive than others, and there are occasions when they need to be re-stated. Enforcement is often through peer pressure, but can also be referred upwards, often by the use of the 'report abuse' or 'flag as inappropriate' features in online scenarios. These norms allow for a shared communal experience, by providing individuals with both implicit and explicit standards and thus facilitating harmonious interaction. Shared experience is an interesting issue - in common with real-world social networks and communities people move from one online community to another. When they join a community, they, perforce, do not share the same set of experience of that community as the long-standing members. In the 'old days' of Internet forums and newsgroups it was generally recommended to 'lurk' – i.e. read posts, but not contribute until you were familiar with the way the group operated. This seems less prevalent in the more modern social networks, where the emphasis is on sharing your own previous experience by asserting your personality
on your profile, blog or any other publishing medium available to you. This is made possible because the mode of use is different - in the social network, everyone gets to decide the agenda and express it in their own ways, whereas the newsgroups and forums were established for a specific topic (or range of topics). In a way, the forum specified the tag and the norms of the community enforced (to a greater or lesser degree) the adherence of content to that tag. This highlights another use of the tag; tagged content can be formatted in the form of a discussion thread. Google mail, for instance, allows the user to do just this, and to view the content filtered by the tags they have chosen to apply to the contents of their mailbox. As tagging becomes more widespread, it is possible to use the tags to follow a topic, just as one would have done on the forums and newsgroups, but from a widely distributed set of sources - and this facilitates each user in becoming more of a prosumer. They can produce whatever content they like, and tag it with words that they think are relevant to it. If it is discovered by others, it can be bookmarked and re-tagged with the terms that the explorers who found it think represent it best - and then reviewers can use it as the basis for their own prĂŠcis, which in turn can be discovered by others. Instead of having to focus on a specific narrow role because of the topic attached to the technology being used, the individual is now able to follow their line of thought and allow the vox populi to determine the merit of their contribution.
Discussion Roles
Figure 8 Illustrating the relationship of roles and identities
It is useful in discussion to expand upon the taxonomy of terms described earlier, which can then be used to analyse and discuss the roles which are adopted in communities. People adopt different roles or positions (Davies, et al., 1990), with attendant behaviours, to suit the context of the environment they are in (Figure 8). In terms of a social network site, for example, several roles can be identified. There are Explorers, who find content which is outside the local network, and post links to it, for their community to see. Then there are Writers who write their own content, a subset of whom are Diarists. Bridging the gap between the Explorers and Writers are the Reviewers, who take existing content and provide analysis of it, to greater or lesser degree. Any role can be in a specific domain, such as an Academic Explorer, who explores issues relating to their field of academic study, or a News Explorer who explores the recent news articles, providing an index for their community.
There are also Socialites, who add content in the context of conversations with other members of the community, and Readers who don’t add content, but consume it. With systems which allow for tagging, there is also a Classifier role, of people who dedicate time to producing meaningful and consistent tags for content. The Ambassador makes connections between multiple services, posting content from one to communities which might find it interesting. Of course, there are also the Troll, and the Flamer, both inherited from usenet communities, and many other roles, and any individual may take on any of these roles at any time within any community. Roles within a community can be composites of several basic role types. Important to online communities in particular, the Technology Steward role (Wenger, et al., 2006) (Wenger, 1998) is to select and configure technology, and provide support for it, within a community, for the benefit of the community, not just because the technology ‘is neat’. This role comprises multiple separate strands, such as the Explorer (in finding new technologies which may support the community), the Reviewer (in analysing the suitability of the technology for the way the group works), a Leader (in driving adoption of technologies which will provide benefit, and gaining consensus to avoid technologies which will not) and Mentor (providing support for the group members as they get to grips with the technology in the context of the group). Also of interest, particularly in communities of practice, the role of Ethnographer, suggested by Smith (Smith, 2007), takes on the task of examining how the technology is used (or not) within the context of the group. This role is effectively a Reviewer, but instead of observing the content produced by the community, they observe the practices and tools and review them. The Ethnographer is in a good position to observe the way the tools change through use. In terms of communities supported by folksonomical tools, these changes can happen in two ways. Modifications can be made to the design of the actual tool (software) which supports the folksonomies which allow the community to express their implicit ontologies. But, more importantly, the folksonomies are tools which the community use, and these are highly dynamic in nature. A folksonomy is not being used as a tool unless it is changing. This realisation brings us to look at the nature of folksonomies through the perspectives of trust networks and Activity Theory.
Activities modify trust Let us consider Amanda and her social network, S (Figure 9). The members of S have their own experience-based opinions of Amanda, based on the content she chooses to share with them and the communication they have with her – based, in fact, on the roles she adopts within the group. As part of their interaction with Amanda and with each other, they make implicit and explicit statements about her, or rather, about her as she is reflected in her roles. Additionally, every member of a network has a model of how much trust they invest in the others in the group in relation to their roles. This trust is essentially their measure of whether the opinions expressed by an individual match up with their own experience, but extends to cover concepts such as the reliability of community members to do what they have committed to, and the level of reasonableness in placing expectations upon an individual to be able to rise to a particular challenge. Even something as simple as Bob re-sharing a link Amanda has posted so that those members of Bob’s social network, T, who aren’t in S can see it, provides a positive affirmation of
Amanda in her role as an explorer. Now there are two cases here. In one, Bob (or the technology he uses) does not attribute the material he shares to Amanda (Figure 9). Consequently, although the shared network S can be aware of the relationship, those in T cannot know that Bob is exploiting his relationship with Amanda. However, if Bob makes it clear that he has re-posted something Amanda drew his attention to (Figure 10), those in both networks will be aware of the relationship between the Bob and Amanda. In the former case, only those people who are in both Amanda and Bob’s social networks will benefit from the expression of trust Bob has made in Amanda’s explorer role. In the latter case, however, anyone who can see Bob’s re-post can learn something about the relationship between Bob and Amanda’s roles within the social network. Also, in the former case, with no attribution, there is the potential for a negative effect on network S’s level of trust in Bob because he does not provide credit, but members of T may gain a false impression of Bob’s exploratory abilities. In the latter case, Bob may not ‘benefit’ from gaining credit at Amanda’s expense, but both benefit appropriately from the actual exchange which has taken place.
Figure 9 Amanda and Bob in the Explorer role, no atrribution by Bob
Figure 10 Amanda and Bob in the Explorer role, with attribution by Bob
The impact of Bob’s decision to re-share the resource Amanda had ‘discovered’ depends largely on three things:
1.
Whether Bob’s re-sharing attributes his source, Amanda
2.
The size of T
3.
The overlap between S and T, S ∩ T
One might think that the level of interest T have in the original resource should be considered as an important factor. However, it probably does not make much of a difference to the level of experience gained by the members of T because, even if they find no interest in the item itself, they learn that Bob sometimes posts things they find dull. If Bob has attributed his post, they can also pick up that Bob sometimes re-posts things which Amanda posted, even if they are dull. If they notice this, they probably start to draw their own conclusions.
It’s not what you know… If Bob attributes Amanda for the ‘discovery’, T all have the opportunity to learn more about the relationship as well as the source information. This allows them to learn about not only the actual subject matter of the original resource, but about Amanda’s interests, about Bob’s interests and about Bob’s interest in Amanda’s interests. If they need to know something later about the topic of the original resource, they now have two more contacts they can use to try to discover more, either in an active dialogue, by asking for advice, or a passive enquiry, by looking at Amanda or Bob’s published material. On the other hand, if Bob does not attribute Amanda, the majority of T only get to learn of the resource, and of Bob’s interest in it, and learn nothing of Amanda. If they find the resource dull, this backfires on Bob, and even if they find it interesting, Bob actually loses out because whilst his network might credit him with the find, they are unable to expand their communal knowledge as quickly as they would otherwise. Only the members of S ∩ T have the potential to benefit as much as possible from Bob’s act of sharing. An additional benefit of attributing in this context is that the act of attribution increases the level of trust T have in Bob, at least to some extent. If he posts without attribution, it may be that he has discovered it himself, or that he is passing on someone else’s find – when he attributes the find to someone else, however, there is a tendency to trust
• The original information more – after all, it has been ‘recommended’ by two or more people already
• Bob’s behaviour in his explorer role, in that if he attributes Amanda, he is likely to attribute me if I post something he wants to share Of course, the network, T is comprised of many individuals, each with their own roles, and their own perspectives of the activity within the network (and indeed, outside it). The activity within the network forms a common environment for its members, but not necessarily a common experience. Attention may be directed elsewhere for any number of reasons, but more importantly each member of the network is actually at the centre of their own social network. The ‘network’ T only exists because Bob is there, at its centre. The connections made by others within T are extremely unlikely to comprise exactly the same ones as those which constitute T itself. There are two cases where T1 and T2 become likely to
be the same – when they are very small, or very close to universal, and in neither case are they useful. When a network is very small, it provides little or no opportunity for synergy to form, and when it is very large it not only suffers from information overload, but also from a lack of classification.
Trust in ontologies One problem is that, as mentioned earlier, the ontology does not directly support the fuzzy nature of trust. This is more the area of belief networks, where inference is done on a probabilistic basis. Most ‘folksonomical tools’, those which support the emergence of folksonomies, only directly support a very binary view – either this resource is tagged with ‘rabbit’ or it is not, either Manu likes seeing the content Gwen tags, or Manu does not. The same is true in social networking sites – either Louise is Manu’s ‘friend’, or not. The fine grain of being able to say “This is 90% likely to be a picture of a rabbit” and “I like to follow the content relating to fluffy mammals posted by Gwen, but not the content related to shotguns” alongside the ability to set rules for friendship – “Geoff is my drinking buddy and can see anything I post, whereas Randolph is my dad, and can see anything I post which is not tagged ‘socialising’” tends to be missing in current systems.
Parallels with folksonomies In folksonomology we are concerned primarily with the emergence of semantics from social behaviours with reference to resources and relationships. However, we can see a close parallel with the emergence of the semantics of the individual. The roles an individual occupies within a community ‘give the individual meaning’, in other words, the digital identities are the meaning of the digital persona in their own specific contexts.
Resources People
Relates to infer
to
apply Tags
Meaning
From
Figure 11 Meaning comes from implicit behavioural relationships as well as explicit acts of creating relationships
Consensual Reality When we need to communicate about something with someone else, we have to have some shared ground. This is the consensual reality that is formed through negotiating about our subjective realities through normal social discourse and through formal interactions. It is the meaning we extract from the relationships we observe, after we have checked with others that our view of the meaning is sufficiently similar to theirs. Consensual reality is the (often implicit) ontology produced through multi-agent collaboration – the folksonomy. The meaning of a symbol, then, is formed subjectively based on our own past experiences, and is an entirely subjective affair, but we try to ensure that it is usable within a social context by iteratively negotiating with the community. This only occurs if we, in conjunction with others, have sufficient interest to negotiate, and consequently the meanings of symbols which are on the periphery of our communal interest will tend to be more vague, whereas those which relate to core concepts will be well defined – in that they will have a better consensus. This carries over to our digital identities. Those roles which are key to the way the community works will have better definition, because they will be the subject of greater levels of negotiation. This suggests that an individual with one or more well defined roles in a community will have more connections, but there is also evidence (Pasupathi, 2001) that they will also tend to define themselves in terms of the community’s shared experience. Indeed, membership of the community is likely to drive their levels of interest (Thoman, et al., 2007), as one’s own level of interest in a subject can be shown to be influenced by the responsiveness of one’s peers when talking about it. In this way, communities have a positive feedback mechanism which encourages them to focus on particular issues. From an educational perspective, this gives good motivation to encourage membership of multiple communities to allow the individual to broaden their range of interests. It is worth exploring further the effects of social network systems on the interests of the users – do we investigate more areas as a result of membership, perhaps losing the focus on specific concerns, or does the potential for disinterest on the part of other community members reduce our own levels of interest, until the network becomes little more than a substitute for a TV entertainment channel?
Conclusion We have argued that personal identity is best represented as a multi-faceted whole, with individual strands of identity going to make up the perception of self. Within the context of a community, the strands of identity are reflected outwards as the roles the individual undertakes. These roles can be emergent, or can be more or less rigidly defined. In either case, the individual will tend to adopt roles which suit them, or for which they perceive that they are best suited given the community members available, given the opportunity to do so. Where roles are emergent, they are described and defined by the folksonomy produced by the community. In more formal systems, they are defined by, or may be described by, an ontology. As online social networking becomes more prevalent, the dynamic nature of community is set to increase, and the Ontology-Folksonomy divide is set to increase. In a culture where transient ad hoc communities are formed, do their jobs, and are dissolved, formal systems become less applicable as the underlying dynamics become more fluid, and methods of analysing and supporting folksonomical community support become more important. Using folksonomies enables the design of automated semantic community role analysis, and allows for the creation of tools to support collaborative working, and communities of practice. Indeed, the close parallels between the role based model and the model of self built on strands of identity suggest that such tools may also help the individual introspect about their role choices and behaviours. Further work using these models of individual and community can be pursued in building software tools for supporting collaborative working and learning and in the future, if the psychological aspects of the modelling are well founded, further folksonomological research may even provide a basis for work towards the development of emergent conscious computing. In the shorter term, however, further work is needed to explore the Ontology-Folksonomy Divide – work on developing tools which can identify roles from behaviours, supporting collaborative working and learning, and building explicit representations of trust networks based on the social interactions of communities on the Internet. Developing the role model with further examples, analysing social networks in terms of roles, behaviours, contributions and relationships, and building tools which can support these, are the next steps in a field where technological practice is often out-pacing academic study.
Glossary Flamer – someone who engages in trading excessive retaliatory insults. Folksonomical – of, or relating to, one or more folksonomies. Folksonomology – the study of folksonomies. Folksonomological – of, or relating to, the study of folksonomies. Folksonomy - representations of domains gathered though processes performed by many people, such as tagging content. Lingo - the language and speech, especially the jargon, slang, or argot, of a particular field, group, or individual. Ontology - a formal specification of a domain, defining and specifying the different classes of individuals that form the domain, the actual individuals and the properties (relationships) of the individuals. Ontology-Folksonomy Divide – the division between ontologies and folksonomies. Prosumer – someone who both consumes and produces a commodity, such as web content, as used by Tapscott in Wikinomics: How Mass Collaboration Changes Everything, Penguin, 2006 Social Network – specifically here, online sites fostering social interactivity. Tag – a word or composite word, possibly with characters other than letters, that an individual chooses to use in relation to material for the purposes of helping them find it again, or in helping others to find it. Tag cloud – a graphical representation of the use of tags in a particular domain, generally using the font size of a tag to represent its frequency of use. Troll – usenet term for someone who deliberately provokes argument, and historically not always considered to be a bad role.
Acknowledgements The authors would like to acknowledge the valuable input from John D. Smith, anonymous reviewers, and that from our team, OdinLab, part of the Ambient and Pervasive Intelligence group of the School of Systems Engineering, University of Reading.
Bibliography Abelson H and Lessing L Digital Identity in Cyberspace [Conference] // White Paper Submitted for 6.805/Law of Cyberspace: Social Protocols.. - 1998. Benkler Y. The Wealth of Networks [Book]. - New Haven and London : Yale University Press, 2006. - ISBN: 0300110561. Blumer H. Symbolic Interactionism: Perspective and Method [Book]. - [s.l.] : University of California Press, ISBN 0520056760, 2004. Boyd S. Are You Ready For Social Software [Online] // Darwin Magazine. - May 2003. - 17 May 2008. http://web.archive.org/web/20050102091600/http://www.darwinmag.com/read/050103/social .html. Clarke R. Computer Matching and Digital Identity [Online] // Proc. Conf. Computers, Freedom & Privacy, San Francisco. - March 1993. - 13 November 2007. http://www.anu.edu.au/people/Roger.Clarke/DV/CFP93.html. Davies B. and Harre R. Positioning: The Discursive Production of Selves [Journal] // Journal for the Theory of Social Behaviour. - 1990. - 1 : Vol. 20. - pp. 43-63. Engeström Yrjö Activity Theory and Individual and Social Transformation [Book Section] // Perspectives on Activity Theory: Learning in Doing: Social, Cognitive & Computational Perspectives / book auth. Engeström Yrjö and al. et. - New York : Cambridge UP., 1999. Galton F Vox Populi [Online] // Nature 75 (1949): 450–51. - 1907. - 13 November 2007. http://tomayko.com/articles/2006/10/27/galtons-ox. Horridge M. [et al.] A Practical Guide To Building OWL Ontologies Using The Protégé OWL Plugin and CO-ODE Tools [Book]. - Manchester : In T. U. O, 2004. Lee C.-S., Jian Z.-W. and Huang L.-K. A Fuzzy Ontology and Its Application to News Summarization [Journal] // IEEE Transactions on Systems, Man, and Cybernetics - Part B: Cybernetics. - 2005. - 5 : Vol. 35. - pp. 856-880.. Marlow C. [et al.] HT06, tagging paper, taxonomy, Flickr, academic article, to read [Conference] // HYPERTEXT '06: Proceedings of the seventeenth conference on Hypertext and hypermedia. - ACM, New York : [s.n.], 2006. - pp. 31-40. DOI:10.1145/1149941.1149949. Merchant G Identity, Social Networks and Online Communication [Journal] // E-Learning. 2006. - 2 : Vol. 3. - pp. 235-243. - DOI: 10.2304/elea.2006.3.2.235. Parry D. A fuzzy ontology for medical document retrieval [Conference] // Second Workshop on Australasian information security, Data Mining and Web Intelligence, and Software Internationalisation. - Dunedin, New Zealand : [s.n.], 2004. Pasupathi M. The Social Construction of the Personal Past and Its Implications for Adult Development [Online] // Psychological Bulletin / ed. Association American Psychological. 2001. - 13 November 2007. -
http://homepage.psy.utexas.edu/homepage/class/Psy394U/Bower/11%20Soc%20Cog%20Per sonality/XX%20Constr%20Life%20Stories/Pasupathi.pdf. - Vol 127, No 5 pp 651-672. DOI: 10.1037/0033-2909.127.5.651. Shirky C. Here Comes Everybody : The Power of Organizing Without Organizations [Book]. - [s.l.] : Penguin Press, 2008. - ISBN: 1594201536. Smith Gene. Tagging: People-Powered Metadata for the Social Web [Book]. - [s.l.] : New Riders, 2008. - ISBN: 0321529170. Smith J.D. A discussion on roles in communities and Communities of Practice. - [s.l.] : Private correspondence, (CPSquare, Learning Alliance), 2007. Stoilos G. [et al.] Fuzzy OWL: Uncertainty and the Semantic Web [Conference] // Int. Workshop of OWL: Experiences and Directions. - [s.l.] : Galway, 2005. Stryker S. and Burke P. J. The Past, Present, and Future of an Identity Theory [Journal] // Social Psychology Quarterly. - 2000. - 4 : Vol. 63. - pp. 284-297. - Special Millenium Issue on the State of Sociological Social Psychology. Surowiecki J. The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations [Book]. - [s.l.] : Little Brown, 2004. - ISBN 0-316-86173-1. Tajfel H. and Turner J. C. The social identity theory of inter-group behavior [Book Section] // Psychology of Intergroup Relations / ed. Austin S. Worchel and L. W.. - Chigago : Nelson-Hall, 1986. Thoman D.B., Sansone C. and Pasupathi M. Talking about interest: exploring the role of social interaction for regulating motivation and the interest experience [Journal] // Journal of Happiness Studies. - Springer Netherlands : [s.n.], 2007. - 3 : Vol. 8. - pp. 335-370. Van Maanen J. E. and Schein E. H. Toward a theory of organizational socialization [Journal] // Research in organisational behavior / ed. Staw B.. - Greenwich, CT : JAI Press, 1979. - Vol. 1. - pp. 209-264. Vander Wal T. Folksonomy Definition and Wikipedia [Online]. - 2 November 2005. - 14 November 2007. - http://www.vanderwal.net/random/entrysel.php?blog=1750. - Blog post. W3C OWL Web Ontology Language Guide [Online] // W3C. - 2004. - 13 November 2007. http://www.w3.org/TR/owl-features/. Weinberger D. Everything is Miscellaneous: The Power of the New Digital Disorder [Book]. - New York : Henry Holt &co, 2007. - ISBN 0-8050-8043-0. Wenger E. Communities of Practice: Learning, Meaning and Identity [Book]. - New York : Cambridge University Press, 1998. - ISBN: 0521663636. Wenger E., Smith J. D. and White N. Definition of Community Technology Steward [Online]. - 2006 December 2006. - 13 November 2007. http://technologyforcommunities.com/. - Blog post.