Proceedings of the 2nd Annual Sessions, Sri Lanka Association for Artificial Intelligence
19th February 2005
Relevance of Buddhist Philosophy to Ontological Modeling in Information Systems and Computer Science
Asoka S. Karunananda and George Rzevski
Abstract-This paper underlines the importance of insightsgained from philosophical studies of ontology forresearch in certain areas of information systemsand computer science. It shows how the understanding of ancient views on ontology mayhelp to resolve some current issues in disciplinessuch as Multi-Agent Systems, Semantic Web andWeb Services. In particular, we single out natural language processing and semantic translation astwo areas of ontological modelling with greatresearch importance where Sri Lankanresearchers could make significant contributions
Intelligent Collaboration among Robotic Agents for Landmine Detection
Vaithilingam Kumarathasa and, Thrishantha Nanayakkara
Abstract-This study is to develop an algorithm for multirobot collaboration for landmine search operation in a typical landmine field found in Sri Lanka. The challenge is to enable robots to work together in an intelligent manner to detect landmines as fast as possible. In Multi-Agent System (MAS) intelligent is considered as some thing emerges through interaction and collaboration among different agents in a swarm. Collaboration among robots is based on a decentralized approach in which robots are based on a set of behaviors; such behaviors are designed to increase global performance and are based in local information and shared information from other team members whenever they are in range of communication. Landmine search method is improved using the prior knowledge about landmine field. Simulation was done to test effectiveness of the algorithm.
Towards Building a Cognitive Vision System for Learning in Behavioral Models from Symbolic Data using Qualitative Spatio-Temporal Relations
D.D.M. D.D.M. Ranasinghe, A.G. Cohn and A. S. Karunananda
Abstract– Research has been carried out to develop a cognitive vision computer system that will be capable of perceiving, reasoning and learning through visual inputs. Our approach is based on the assumption that robust qualitative spatio-temporal relations extracted from visual data can be used to successfully implement cognitive vision systems that behave like humans with reasoning and learning abilities. In our ongoing research, placing covers on a dinner table has been analysed and have been identified some basic robust qualitative spatiotemporal relations such as rightof, leftof, frontof, etc. Prolog has been used for implementation of analysis of spatio-temporal relations, while extraction of relevant rules from relations has been implemented with the help from Progol, which is a many sorted language for inductive logic-programming that implements learning by examples. The final goal of this project is to make a computer-based cognitive vision system capable of learning from more comprehensive set of examples and carry out complex reasoning on spatiotemporal visual data to learn behavioural models
Using Fuzzy logic for automating knowledge acquisition
D.S. Kalana Mendis, Asoka S. Karunananda and U. Samrathunga
Abstract– Knowledge acquisition is a key phase in construction of expert systems. This process is very much dependent on the nature of the domain knowledge. In particular, knowledge acquisition is a tedious task when modeling domains with tacit knowledge. Despite fuzzy logic has been used for knowledge acquisition in such domains, a large portion of the process is manually operated. This paper presents an approach to automated knowledge acquisition using fuzzy logic for the domains with tacit knowledge. The novel approach allows the user or developer to directly interact with the system and enter tacit form of knowledge Acquisition of tacit knowledge is supported through questionnaire emulating the role of an interview of domain expert by the knowledge engineer. Tacit knowledge acquired through this session will be analyzed by statistical technique of principle component analysis to reveal the available dependencies in the knowledge acquired. The principle components generated will be transferred to fuzzy logic module for automatic construction of the fuzzy membership functions. Further, in a usual manner, system can acquired fuzzy rules relevant to manipulation of domain knowledge. Thus, collectively, the approach is consisted of principles component analyzer, fuzzy logic module and a fuzzy expert system. The approach has been developed to be able to connect with a standard expert system shell. Currently, it has been integrated with FLEX expert system shell. The development has been done using Visual basic and the system runs on Windows platform. The approach has been applied to many domains include Ayurvedic domain of classification of individuals. It has shown 78%. accuracy in using the tacit knowledge for reasoning in the relevant domain. Performances were very close to handling tacit knowledge by the human expert in tacit domain.
English to Sinhalese Bilingual Web Client for accessing the Semantic Web
Asoka S. Karunananda and Budditha Hettige
Abstract-Bilingualism has been recognised as a means for using the power of mother tongue for comprehending materials in a second language. We have been developing an English to Sinhalese bilingual translator as an Expert system, known as BEES, which can operate on an ordinary Web Client enabling the access to World Wide Web in a bilingual manner. The current system is capable of translating simple web pages appeared in English so as to generate bilingual displays of the web pages containing a certain percentage of English and Sinhalese words, depending on the user profile. BEES has been incrementally tested and currently we are working for improving BEES to handle semantics of bilingual translation towards a novel approach to ontological modelling for the Semantic Web.
An Agent Based Visual Language for Virtual Worlds
J. Fernando, D.D. Karunaratna and G.D.S.P. Wimalaratne
Abstract-It is important to allow creation of dynamic virtual worlds, which interact to user inputs. Currently, there is no simple standard method to model interactions and dynamic behavior in virtual worlds. This paper proposes an agent based visual programming language for modeling behavior in virtual worlds that facilitates creation of dynamic, autonomous virtual worlds. The visual language represents virtual world object behaviors based on agents and a hybrid of finite state machines, fuzzy state machines and petrinets, combined with message passing, dataproperties and persistence. The agent based visual language simplifies the process of modeling highlevel behavior of virtual worlds
“Gene Doctor” – System that diagnoses genetic diseases by analyzing, DNA sequences through use of “artificial neural networks”
A.J. Peris and A. Wickramasinghe
Abstract-Humans are at times infected with genetic diseases caused by a mutation in DNA. Getting a DNA test done is a very strenuous, timeconsuming and costly task, especially due to the complexity, and the unpredictable nature of mutations. Therefore, once a human get infected with a genetic disease, he or she has to go through a lot of hassles before getting a good diagnosis for his or her sickness. Artificial intelligence can be used to solve the above-mentioned problem. There are many artificial neural network techniques that could be used to solve this kind of problems. But, artificial neural networks would be the best artificial intelligence technique to solve this particular problem due to the complexity and unpredictable nature of the DNA sequence.”Gene Doctor” is a computer system that diagnoses genetic diseases through analysis of DNA sequences. The analyzing of the DNA sequence is done through an artificial neural network. The artificial neural network used in ” gene doctor” is a three layer neural network with Backpropagation as its Learning Rule, for the working prototype of “Gene Doctor”. The prototype is currently trained to detect four most common genetic diseases with an accuracy of 78%.”Gene Doctor” also includes some other features such as Gene Therapy Finder, Graphical Sequence Simulator, Detail DNA Report, and Automated Translation & Transcription. The above features are helpful to doctors and scientists and lab assistance that work in microbiology and DNA research field.
An Intelligent System to Classify Varieties of Wood
B. Modasia and M.A.De Silva
Abstract-Grading by appearance is common in the wood industry and mostly carried out manually. The classification based on anatomical structure is indeed a good technique for wood identification. However, the lack of expertise in the said field has motivated the author to develop of a computerized system in order to achieve this task. The use of computers in the wood industry has now become an established norm in many developed countries. With the advent of internet, computers are readily used for the retrieval of microscopic images of wood, especially for research purposes, comparisons etc. Since wood identification is done manually, use of computers would be an ideal approach to help increase the accuracy of the identification process. The overall approach describes the development of “Intelligent Wood Grain Processor” that provides an edge over the cumbersome traditional manual identification methodologies. Use of Digital Image processing and Artificial Neural Networks helped the author develop this system, with an accuracy of 75%, to identify microscopic images of wood.