Proceedings of the 7th Annual Sessions, Sri Lanka Association for Artificial Intelligence

12 November 2010

Ontology – What and Why?

Damitha D Karunaratna

Abstract: Today ontologies are very popular in many areas of computer science including
Knowledge Management and are being used in a variety of applications. There are many definitions
for the term “Ontology”. Also, different applications use different contents and structures in building
ontologies. This paper gives a brief account on why ontologies are needed, how they are defined and
used in knowledge management.



Varanageema: A Theoretical basics for English to Sinhala Machine Translation

B. Hettige, A. S. Karunananda

Abstract – Grammatically correct target language word generation is a challenging task in any machine translation. To achieve this challenge, most of the machine translation systems used theoretical basics with rule-based approaches. This paper reports the theoretical basics of the English to Sinhala machine translation named as
concept of Varanegeema. The English to Sinhala machine translation system has been developed as
a multi agent system and it provides grammatically correct Sinhala translation for given English text. This system uses the concept of Varanegeema in Sinhala language as theoretical basics for English to Sinhala machine translation. This Varanegeema concept also reduces the Sinhala word availability in the Sinhala dictionary
and gives the way to generate grammatically correct Sinhala words. The translation system has been successfully tested through limited domain and the obtained results that we consider very satisfactory. Details of the concept of Varanegeema, design, implementation and evaluation of the English to Sinhala machine translation are given in this paper.



Usage of Agent Technology for Distributed Medical Services Support

B. S. T. Karunarathna, A. S. Karunananda 

Abstract: Typical health systems don’t adequately address the complexities in the modern health context in an efficient manner. Necessarily dynamic and distributed nature of the health environment requires being adaptive to the changes taking newer decisions accordingly to co-evolve with the situation, which best would be enable using the common space provide in the Multi Agent Technology to build effective interaction among associated parties. Therefore this paper presents a better approach in delivering assistance for the health service seekers enabling enhanced accessibility to medical services streamlining the health care process adapting emerging technique of Multi Agent Technology which best would be deal with the complex nature of the health context enabling the self adaptation necessarily with the dynamically changing instances. Mainly system follows the message-request-resource-ontological model which comprises of three main agent modules respectively for the channeling, medical tests and pharmacy services which activates necessarily based on the incoming requests and derives necessary conclusions through effective communication, coordination and negotiation. Finally selecting suitable parties covering the health domain successful evaluation was carried out to access the level of success gained at the end



SAEE – Swarm of Agents for English Editing
A. Kiribathgoda, A. S. Karunananda
Abstract In this knowledge economy, writing comprehensively is a chief value-producing activity. Natural Language editing in an automated environment is a complex task and due to this complexity, current automated tools face the limitation of perfectly editing a given write up considering both syntactic and semantic accuracy. Therefore, there is a need of evolutionary, adaptive, autonomous and dynamic approach for natural language editing. As the solution ‘A Swarm of Agents for English Editing’ (SAEE) is a multi agent based approach for evolutionary English editing. This system comprises a swarm of agents with spell checking agents, noun agents, verb agents, determiner agents, style agents and manager agent where each agent activates upon a request, executes and kills them at the end of the task. Unlike traditional editing software this system is based on negotiations among different agents rather than instruction driven. In this system English editing task involves interaction among different agents who take care of different aspects of a sentence structure such as noun, verb and determiners. These agents together with the manager agent negotiate with each others to generate an agreeable output. A common message space is used to enable the communication and thus negotiations among different agents in the swarm to come up with a commonly agreeable edited text.



iAgri – An Agent Based Agricultural Information Provision
W. C. M. Kumarasiri, Asoka S.Karunananda

Abstract – Agriculture has become one of the most important areas of the world with the current food crisis. But, still there is lack of technological aid that can be assisted to increase the efficiency and effectiveness of the agricultural sector. To properly manage the whole process, there is a need for expertise knowledge and a proper communication channel between stakeholders. Nevertheless there is no fully fledged mechanism between farmers and agricultural experts to interact and share the knowledge effectively. This system is an effort to solve the problem of deriving information from scattered and distributed sources. As the solution, an Agricultural Value Chain Management system based on agent technology has been implemented to provide common space for improving the transparency of information by centralizing disseminated agricultural expert information whenever it’s needed to the farmers. The system divides agricultural value chain in to 6 stages called Deciding, Seeding, Preparing and Planting, Growing, Harvesting and Selling. The necessary information needs are identified at the each stage by the system and the information is derived from the various agricultural information sources using agents. Then the most suitable information is provided according to the crop and the value chain stage of the crop via a multi agent system.



Resilience Modeling of Value Chains

G. I. Premarathne, A. S. Karunananda
Abstract – Resilience Modelling of Value Chains involves changing behaviours of the occupied entities dynamically, hence we need an adaptive, flexible and responsive solution to achieve this scenario. Multi Agent Systems could provide a common space for enabling the interaction among Value chain entities. This paper describes the development of an agent-based software system for assisting in decision making regarding value chain management and to achieve resilience modelling of value chains in the milk powder industry in Sri Lanka. Such a system can be applied to different types of industries with some domain specific modifications. The core architecture is built around the concept of Multi Agent Systems since Agent Technology has been identified most suitable for modelling complex scenarios.



Agent based autonomous Graphic Design Tool “PICTURE IT”

H. B. Walisundara, A. S. Karunananda

Abstract: In the current world all the activities are done by the use of communicating through very precise as well as an unambiguous Medias. That’s why people try to convey their massages by displaying it using posters, brochures as well as by leaflets. To convey their massage very precisely from such kind of Media, it should be properly design by people who has expertise knowledge about graphics as well as the design principals. So if a person who doesn’t have such kind of expertise knowledge wants to design a poster, cover page of a book, he should hire that kind of expert. Sometimes it may be expensive for normal person to get his work to be done.  PICTURE IT is a designing tool that solves the problems arise in designing any paper work and maximizes the efficiency and preciseness in to a greater extend which anyone needed the most. This tool is implemented using multi agent technology which provides an intelligence that gives the designer a convenient and user friendly environment and builds the confidence as well. It is different from other applications that had been implemented for design purpose. That is it consists of some kind of autonomy that helps the user to design his or her imagination in to a successful media of convey the message. This tool has the ability to self organize by itself which provides answers to the many questions in designing a poster, cover page, letter heads, brochure etc.



Recruitment Support System with CV Classifier Expert System

W N R De Alwis, H U W Ratnayake

Abstract: Recruitment Support System (with CV classifier) is a web based system in which job seekers can register online and apply for a job and attend the initial selection test given by the company. It involves uploading CV, classifying CVs, and recommending job categories that the CV is most suitable. The system is developed as a web based environment to provide a user friendly, online and multi-user access environment. An Expert system is used for classification of CVs. The rule base was implemented using XML and the developed prototype successfully gives a suitable job category for a given employee data.