Combining ontology and folksonomy: An Integrated Approach to Knowledge Representation

1. Combining ontology and folksonomy:An Integrated Approach to Knowledge Representation Atefeh Sharif PhD student of library and Information Science Ferdowsi University of Mashhad, Iran (Islamic Republic) at_sh91@stu-mail.um.ac.ir atefehsharif@gmail.com 19-20 Aug, 2009- Florence

2. Summary • Introduction • Folksonomy: Collaborative tagging • Small Universe & three problems • Ontology: formal specification of a shared conceptualization • Bridging between two ends of a range • An Integrated Approach • Modeling • tow sub-models • Conclusions

3. Introduction • Technological advances • change in KnowledgeRepresentation and Retrieval approaches • two approaches and a shift • social, flexible, dynamic, lightweight, user-dependent content creation and classification e.g.. collaborative tagging in folksonomies • formal, solid, static, system-dependent model of knowledge e.g..thesaurus and ontologies

4. Folksonomy: Collaborative tagging • Web 2.0 philosophy • Annotating resources with a set of words (tags) • According to the users’ needs • Without relying on a controlled vocabulary or a previously defined structure • Results in • User dependent, personal and also shared sets of tags • Represents users’ • interests, knowledge, interpretations, social or cultural background • Perception of the world Specific domain of interest

5. Small Universe & three problems • Easily create, rename, group, split, merge and delete tags • Problems with • Ambiguity in meaning (polysemy) • Tag variation (synonymy) • Flat organization • Leads to • reduction of search capacities • limited navigation • lack of precision There are some problems with the first approach

6. Ontology:formal specification of a shared conceptualization • Knowledge representation based on conceptualization of entities and relations in a Possible world • Features: • Provides machine understandable information • Requires consensual agreement on its contents form community members • expensive in creation & difficult in maintenance (mainly in changing environments) • to be effective, they need to change possibly as fast as the parts of the world they describe There are some problems with the second approach too!

7. Bridging between two ends of a range • Search both within and across a folksonomy based system is an open problem but content retrieval can be further improved by making relations between tags explicit • It would help to continue the searching process and to do query-refinement in folksonomies • high effort needed to develop and maintain sophisticated ontologies and • ontology users are not involved in the development process sufficient involvement of users in the construction of ontologies can be happened through the use of collaborative environments it is possible to get the both technologies complement each other an Integrated approach

8. An Integrated Approach • use the benefits of both ontologies and folksonomies to complete the process of knowledge acquisition and representation on web It is possible to have • the flexibility of use and implementation of folksnomies • in fostering collaboration among users • improving the process of ontology engineering • the structured model of knowledge in ontologies • to overcome the current limitations – poor searching and navigation capabilities- in folksonomies

9. An Integrated Approach (cont.) • It seems that no scholarly paper have built conceptual model of any integrated system to show • how such systems will work and • how users will treat them • We are neither going to • propose a new approach in search expansion • Nor to investigate a set of tags to extract concept and relations • We just intend to deepen the understanding of an integrated approach in a knowledge representation system

10. Modeling • basic Assumptions: • Folksonomies hold more semantic value than keywords extracted using statistical approaches • Implicit knowledge can be derived by means of statistical analysis of the annotations combined with pragmatic information provided by the semantic web and additional clues given by external resources • the proposed model is split into twosub-models • knowledge (or better say information) acquisition and representation • knowledge discovery (search & navigation)

11. Repository Folksonomy DB Ontology 1 Ontology 2 Controlled tagging Uncontrolled tagging First sub-model Knowledge acquisition and representation Candidate tag/concept Candidate tags/concepts Candidate tag refinement Candidate tag refinement No results Recommended tag/concept Existing tags (clustered) existing concepts (visualized) Tag assignment New tag/concept suggestion Confirm tag/concept Automated Concept and relation extraction Ontology evolution Mapping tags into concepts/instances/ properties of ontology Expert’s confirmation Repository update

12. Repositories Folksonomy DB Ontology 1 Ontology 2 Second sub-model Knowledge Discovery (search & navigation) Search a query Browse Query Refinement Alphabetic list of tags, concepts, Actors, groups Query Refinement Visualized view: tag and actor cloud, structured concepts Actors, groups Clustered and/or visualized tag/concept No results Selection/end

13. Conclusions • Enhanced information retrieval could be the result of any ontology-based folksonomy system • valuable knowledge available in folksonomies could allow keeping online ontologies up to date • first sub-model shows • how one user can enrich tagging process by the available ontologiesand • how collaborative systems support dynamic ontology evolution • second sub-model indicates • the indirect effects of ontologies linked/embedded to the system to improve information retrieval • These two sub-models are simplified to some extend

14. Conclusions (cont.) • While we are at the beginning of realizing benefits of such integrated systems, there is a clear magic as we see semantics emerge from the individual actions of a community during their real activities as a member in social networks • It is needed to improve and modify these two proposed sub-models So • we need tonegotiate with professionals from different related fields, such as Computer science and Semantics • Library and Information specialists in cooperation with other specialists can help speed up the movements towards the semantic web

15. Thank you Emam Ali said: THE WORLD IS YOUR MOST ELOQUENT COUNSELLOR , IF YOU ARE SUSCEPTIBLE TO COUNSEL