Title: Discovering Causal Dependencies in Mobile Context-Aware Recommenders

Speaker: Mr. Yap Ghim Eng (Phd Student, ER Lab)

Abstract:

Mobile context-aware recommender systems face unique challenges in acquiring context. Resource limitations make minimizing context acquisition a practical need, while the uncertainty inherent to the mobile environment makes missing context values a major concern. This paper introduces a scalable mechanism based on Bayesian network learning in a tiered context model to overcome both of these challenges. Extensive experiments on a restaurant recommender system showed that our mechanism can accurately discover causal dependencies among context, thereby enabling the effective identification of the minimal set of important context for a specific user and task, as well as providing highly accurate recommendations even when context values are missing.

Click here for presentation slides

Date: 05 Apr 06, 4pm

Title: A Goal Net Based Agent Development Environment and Its Applications

Speaker: Mr. Li DongTao (Master Student, ER Lab)

Abstract:

Agent technology represents a new software paradigm. The autonomous, intelligent and goal-oriented characteristics make agents a promising solution for the next generation of software. However, despite of the research effort on agent modeling, there is still a lack of widespread deployment of agent systems. The major reason is that the research narrowing the gap between agent mental state design and agent implementation is rare.

To bridge this gap, we propose a Multi-Agent Development Environment (MADE) based on Goal Net Agent Development Methodology. MADE works as a supplementary tool for Goal Net, to simplify agent development work by providing comprehensive guidance to developers and translate the agent mental state design into agent implementations. The environment consists of two major parts: Goal Net Designer and Agent Creator. Goal Net Designer provides a graphical user interface for designing Goal Net; according to the Goal Net design, Agent Creator is able to create an agent which is minded with the designed goal net. The low-level implementations details become transparent to agent developers. In this talk, we present the development details of MADE and several cases studies of using MADE to develop agent systems.

Date: 30 Nov 05, 4pm

Title: Cooperative Cognitive Agents and Reinforcement Learning in Pursuit Game

Speaker: Mr. Xiao Dan (Phd Student, ER Lab)

Abstract:

This paper illustrates how a self-organizing cognitive architecture, known as TD-FALCON, can learn to function and cooperate in a dynamic environment. TD-FALCON learns the value functions of the state-action space estimated through a temporal difference (TD) method. The learned value functions are then used to determine the optimal actions based on an action selection policy. To tackle a multi-agent predator/prey pursuit task, we develop a cooperative strategy using a high-level compressed state representation and a hybrid reward function. Experiments show that TD-FALCON agent teams operating with the proposed cooperative strategy produce superior performance with a high level of efficiency and scalability.

Click here for presentation slides

Date: 26 Oct 05, 4pm

Title: Mining Ontological Information From Domain-Specific Text Documents

Speaker: Mr. Jiang Xing (Phd Student, ER Lab)

Abstract:

Traditional text mining systems employ shallow parsing techniques and focus on concept and taxonomic relation extraction. In this talk, we will present a system called CRCTOL for mining rich semantic knowledge in the form of ontology from domain-specific text documents. By using a full text parsing technique and incorporating both statistical and lexico-syntactic methods, the knowledge extracted by our system is more concise and contains a richer semantics compared with alternative systems. We conduct a case study wherein CRCTOL extracts ontological knowledge, specifically key concepts and semantic relations, from a terrorism domain text collection. Quantitative evaluation, by comparing with a state-of-the-art ontology learning system known as Text-To-Onto, has shown that CRCTOL produces much better precision and recall for both concept and relation extraction, especially from sentences with complex structures.

Click here for presentation slides

Date: 28 Sep 05, 4pm

Title: On Efficiently Mining Generalized Association Rule from Large RDF Metadata Collection

Speaker: Mr. Jiang Tao (Phd Student, ER Lab)

Abstract:

Semantic Web proposed by Tim Berners-Lee is a vision of next generation web which will be flourished by machine understandable documents. Machine-processible semantic metadata will be widely used and generated for describing web resources with explicit semantics. Resource Description Framework (RDF), developed by W3C, is a language specification for representing semantic metadata. Due to the continual popularity of the semantic web, in a foreseeable future, there will be a sizeable amount of RDF-based content available on the web. Just as data mining and web mining is critical for extracting useful information from today’s World Wide Web, discovering knowledge from machine understandable web contents will become an essential part for tomorrow’s usage of semantic web. In this talk, we present our work of mining generalized association rules from large RDF metadata collection. Based on the features of RDF relation lattice, several optimization strategies are proposed. Experiments conducted on synthetic datasets and a real-world FOAF (friend-of-a-friend) dataset showed that our proposed methods can efficiently find useful rules in large RDF databases.

Click here for presentation slides