D2.1: Inventory of Topics and Clusters

Title: THE WIKI CONCEPTUALISATION OF THE IDENTITY DOMAIN IN FIDIS

The WIKI conceptualisation of the Identity domain in FIDIS

Defining “the perfect ontology” of a domain as complex and evolving as the Identity domain represents a major (and probably impossible) undertaking that the FIDIS project does not pretend to achieve: 

First, because there exists many different possible categorisations of terms, each of them more adequate for the use in a particular context; Second, because the domain “Identity” is very dynamic and continuously evolving (new issues are emerging), and cannot be easily “institutionalised” yet; Third, and as a consequence, because the elaboration of a “perfect ontology” may not be a desirable goal: the objective of this ontological work is not so much to impose on a community a particular vision of the meaning of identity, but rather to give this community a tool (the shared vocabulary) that will help it to exchange and share ideas, and to collaborate in the creation of new knowledge and meaning (Wenger Etienne, 1999);  Fourth, because this represents an unfeasible task: ontology building is still more of an art than an engineering discipline, and requires considerable time and effort.

The FIDIS “Identity” ontology has therefore been considered from the outset as a tool for the community, formed by the FIDIS network, to develop a shared understanding of the domain. The value of this “Identity” Ontology construction is therefore not only seen by FIDIS as simply its first output (the set of definition of terms), but also a way of providing a mechanism that helps this community to developed its own – dynamic and diverse - identity. 

The FIDIS “Identity” Ontology is meant from the very beginning as a “living entity”, constantly engaging the FIDIS members in some interaction, and continuously integrating the new findings of the “identity” domain. This collaborative process involves all the members of the FIDIS community in proposing terms, categorisations, definitions and illustrative examples. 

The use of a WIKI appears to represent a tool of choice for supporting this vision of continuous process of conceptualisation, and was therefore adopted. 

In the following chapter, we are going to present the different process of building the Ontology that will be “captured” into the FIDIS Identity WIKI (the description of process of building the textual conceptualisation of chapter will not be presented, because it only consisted in the well know editorial work).

The construction process of the FIDIS Identity conceptualisation

In order to create the FIDIS Ontology, a bottom-up, constructivist, iterative and collaborative approach was adopted. This choice has been preferred in order to better manage the relative novelty of the Identity field, which is not yet stabilised, to allow more flexibility (and rearrangement), to facilitate the collaborative process and the progressive emergence of meaning. An alternative approach (that has not been adopted) would have consisted in a more linear process putting a stronger and earlier focus on the formalisation of the structure (definition of a “classification” and of the conceptual relations between the terms), and the utilisation of this structure to “organise” the different terms (seen as instances of classes, connected to one another via well defined semantic relations). 

Besides, this approach was supported by the use of a Wiki, provided as part of the FIDIS information infrastructure. A Wiki offers much flexibility, and is well adapted for concurrent and collaborative activities. 

As we have indicated previously, the construction process of the FIDIS ontology consists of iterations of the following operations: 

1. Identification of terms, concepts, etc. 

2. Categorisation of these terms 

3. Definition of their semantic 

4. Illustration (situating them with examples) 

5. Some tentative formalisation of the conceptual relations 

The Identification of terms, concepts, etc… (corpus analysis)

The identification process of terms and concepts has consisted of: 

1. The extraction of terms from this collection of Internet resources

2. The extraction from more formal identity domain documents (such as (Hansen and Pfitzmann, 2004), or the FIDIS deliverable elaborated in the project). 

3. A continuous process of collecting terms from the FIDIS members. 

The first operation (of course very incomplete) of collection of Internet resources related to the concept of identity, was aimed at making a first state of the art description of this domain so as to get a first idea of the main identity issues, but also to collect a set of reference materials that could be used later to illustrate a particular identity issue. Examples of the categories of Internet resources include: sources of information, events, journal and magazines, articles, actors and categories of actors in this domain (companies, laboratories, etc.), projects and initiatives (research projects, consortium), etc. This collection of resources helped in particular to identify the terms, concepts, mechanisms and issues that are the more frequently found related to identity. Examples includes items such as phishing (criminal activity relying on the use of a fraudulent websites mimicking an official web site, which is designed to fool recipients), RFID tagging (which consists in the use of a rfid device to add identification attributes to an object), or adware (software pushing advertisements and that typically spy the activities of the users).

The extraction from more formal identity domain documents (which can be considered by themselves as text-based specification of conceptualisation) help to identify, and to find definition of the more elaborated identity concepts. For instance (Hansen and Pfitzmann, 2004) defines very thoroughly a set of identity concept such as: Anonymity, Unlinkability, Anonymity, or Pseudonymity. These concepts that can directly be imported and become part of the FIDIS conceptualisation (with the appropriate acknowledgment of the origin of the work done). This extraction process can be considered as even more important in the case of the use of the documents deliverable generated in the FIDIS project, since it provides the means to link all the different stated definition of identity that are formed in the different workpackages. 

Note: the use of the FIDIS deliverable documents was very limited in the first version of the specification of conceptualisation generated by the workpackage WP2. Indeed, few of these FIDIS deliverables were available at the time of the elaboration of  del 2.1., and besides the objective of this first version was principally oriented in laying-up solid foundations (principles and methodologies) for the elaboration of the FIDIS conceptualisation.

The categorisation

The categorisation in FIDIS has consisted in some process of aggregation of identity terms that had been identified (see “ ” for the result of categorisation). The idea was to be able to manipulate the different terms in a way that would brought a little bit more sense than would do an alphabetical organisation. For instance this categorisation helps distinguishing descriptive perspective of identity (presented in “ ”), a processes perspective of identity (presented in “ ”), a tool perspective (see “ ”) or domains of application (see “ ”). This process of categorisation was mainly empirical, and did not rely of well defined methods of categorising a domain. This approach had the advantage of simplicity and did not require some sophisticate approaches or theories that are not really available in the context of this work. Automated clustering techniques represent a good illustration of more sophisticated mechanism of classification. They can be used to automate the acquisition of taxonomies or concept hierarchies from a text corpus (Cimiano, Hotho, and Staab, 2005) or the construction of Ontologies (Bisson, N´edellec, and Canamero, 2000).

Without having to go to the extreme of automated categorisation, we are aware that the use of more rigorous categorisations methods in FIDIS would be desirable, and will be investigated in the future. 

The definition of the semantic

The definition of the semantic that was adopted in FIDIS and consisted in the very traditional definition of the term that is traditionally used inside WIKIs. 

It has to be mentioned that the definitions of the terms have tried to conform to some templates (more details can be found “ ”) that provided some way to systematise and homogenise the representation. For instance, a term (entry in the WIKI) was seen as an object possessing attributes such as: textual definition related terms, issues, application domains, etc. (different categories of term objects exist).

Practically, this structure will only be enforced by guidelines and good practices, since these different terms will only be entered using text description in the WIKI system. 

The illustration (Situating the concept in concrete contexts)

“Exemples or illustrations” were / are encouraged for each WIKI entry. For instance, the term anonymity may be illustrated by a set of examples of situations involving anonymity, and helping to illustrate the different anonymity issues. 

Practically, this illustration may appear as an extension of the semantic definition that we have mentioned previously, and will only consist in adding an “example attribute”. 

Note: the cases stories and scenario of Del 2.2 represent typical examples that could be reference in the future in this section. 

Formalisation

This step consists in the more explicit definition of the templates that we have mentioned, and aim at homogenising the definitions of the different terms and their relation with one another (and for instance, identifying the conceptual relation between the different terms). Beside some processes of consolidation and “cleaning” consisting in rewriting the description of the concepts can be done (this is particularly relevant if several authors have contributed independently to description of a single term).