Ching Man Au Yeung

Contextualising tags in collaborative tagging systems

Collaborative tagging systems are now popular tools for organising and sharing information on the Web. While collaborative tagging offers many advantages over the use of controlled vocabularies, they also suffer from problems such as the existence of polysemous tags. We investigate how the different contexts in which individual tags are used can be revealed automatically without consulting any external resources. We consider several different network representations of tags and documents, and apply a graph clustering algorithm on these networks to obtain groups of tags or documents corresponding to the different meanings of an ambiguous tag. Our experiments show that networks which explicitly take the social context into account are more likely to give a better picture of the semantics of a tag.

Tag Meaning Disambiguation through Analysis of Tripartite Structure of Folksonomies

Collaborative tagging systems are becoming very popular recently. Web users use freely-chosen tags to describe shared resources, resulting in a folksonomy One problem of folksonomies is that tags which appear in the same form may carry multiple meanings and represent different concepts. As this kind of tags are ambiguous, the precisions in both description and retrieval of the shared resources are reduced. We attempt to develop effective methods to disambiguate tags by studying the tripartite structure of folksonomies. This paper describes the network analysis techniques that we employ to discover clusters of nodes in networks and the algorithm for tag disambiguation. Experiments show that the method is very effective in performing the task.

A k-Nearest-Neighbour Method for Classifying Web Search Results with Data in Folksonomies

Traditional Web search engines mostly adopt a keyword-based approach. When the keyword submitted by the user is ambiguous, search result usually consists of documents related to various meanings of the keyword, while the user is probably interested in only one of them. In this paper we attempt to provide a solution to this problem using a k-nearest-neighbour approach to classify documents returned by a search engine, by building classifiers using data collected from collaborative tagging systems. Experiments on search results returned by Google show that our method is able to classify the documents returned with high precision.

User-induced links in collaborative tagging systems

Collaborative tagging systems allow users to use tags to describe their favourite online documents. Two documents that are maintained in the collection of the same user and/or assigned similar sets of tags can be considered as related from the perspective of the user, even though they may not be connected by hyperlinks. We call this kind of implicit relations user-induced links between documents. We consider two methods of identifying user-induced links in collaborative tagging, and compare these links with existing hyperlinks on the Web. Our analyses show that user-induced links have great potentials to enrich the existing link structure of the Web. We also propose to use these links as a basis for predicting how documents would be tagged. Our experiments show that they achieve much higher accuracy than existing hyperlinks. This study suggests that by studying the collective behaviour of users we are able to enhance navigation and organisation of Web documents.

Distributed human computation framework for linked data co-reference resolution

Distributed Human Computation (DHC) is used to solve computational problems by incorporating the collaborative effort of a large number of humans. It is also a solution to AI-complete problems such as natural language processing. The Semantic Web with its root in AI has many research problems that are considered as AI-complete. E.g. co-reference resolution, which involves determining whether different URIs refer to the same entity, is a significant hurdle to overcome in the realisation of large-scale Semantic Web applications. In this paper, we propose a framework for building a DHC system on top of the Linked Data Cloud to solve various computational problems. To demonstrate the concept, we are focusing on handling the co-reference resolution when integrating distributed datasets. Traditionally machine-learning algorithms are used as a solution for this but they are often computationally expensive, error-prone and do not scale. We designed a DHC system named iamResearcher, which solves the scientific publication author identity coreference problem when integrating distributed bibliographic datasets. In our system, we aggregated 6 million bibliographic data from various publication repositories. Users can sign up to the system to audit and align their own publications, thus solving the co-reference problem in a distributed manner. The aggregated results are dereferenceable in the Open Linked Data Cloud.

Discovering and Modelling Multiple Interests of Users in Collaborative Tagging Systems

We analyse data obtained from several collaborative tagging systems and discover that user interests can be very diverse. Traditional methods for representing interests of users are usually not able to reflect such diversity. We propose a method to construct user profiles of multiple interests using data in a collaborative tagging system. Our evaluation suggests that the proposed method is able to generate user profiles which reflect the diversity of user interests and can be used to help provide more focused recommendation.

Collective User Behaviour and Tag Contextualisation in Folksonomies

Collaborative tagging systems have emerged in recent years to become popular tools for organising information on the Web. While collaborative tagging offers many advantages, they also suffer from several limitations, with a major one being the existence of ambiguous tags. To understand what an ambiguous tag is intended to mean, we need to know the contexts in which it is used. Instead of using common large scale clustering techniques on folksonomies, we believe tags can be better contextualised by the social contexts in which they are used. We propose a method to reveal semantics of ambiguous tags by studying the collective user behaviour in a tagging system. In this paper we describe our proposal and some results of our preliminary experiments. We also discuss the significance of the work and how it can be evaluated.