Rule Based Multi Agent System To Reduce Malnutrition In Children

E-government is the use of Information Technology to free movement of information to overcome the physical bounds of traditional paper and physical based systems. Therefore the use of Information Technology can enhance the access and the delivery of government services to benefit citizens, business partners and employees in a country. When considering the e-government initiatives in Sri Lanka there are a lack of Information Technology related projects in health sector. According to a recent study child malnutrition continues to be a serious health concern in post independent Sri Lanka. Therefore it is important to address the causes of under nutrition, as it will impede productivity, economic growth and poverty reduction in a country. During the recent past, Sri Lanka has improved in nutrition, underweight rates declined by almost 25%, long term under nutrition declined by about 75%.Therefore this research mainly focus on the current situation of e-government and malnutrition status in Sri Lanka and how we can use Agent Technology to detect malnutrition and distribute nutrition knowledge among the citizens. As Agent Technology has the ability to provide an autonomous solution for communication and sharing of knowledge among different sources. As well as Multi Agent System is a collection of software agents that work, by communicating with each other. They sometimes cooperate or compete with each other or a combination of corporation and competition, yet with a common goal to be achieved. Furthermore it will use rule based classification technique in data mining in order to detect malnutrition among children under five. As data mining is a branch of computer science and artificial intelligence and is a process of extracting patterns from data. Data mining is seen as an increasingly important tool by modern business to transform data into business intelligence giving an informational advantage. It is currently used in a wide range of profiling practices, such as marketing, surveillance, fraud detection, and scientific discovery. A rule based classifier is a technique for classifying records using a collection of "if... then..." rules. Therefore it will extract the rules from the data set and try to find out whether there is any connection between the numbers of rules used to make the final decision with the optimality of the final decision. The Java Agent Development Framework (JADE) software framework facilitates the development of interoperable intelligent multi-agent systems that is used by a heterogeneous community of users as a tool for researchers as well as for real world applications. Hence in this thesis it presents the design and architecture of a multi-agent system to reduce malnutrition (MASRM) which is developed on top of the JADE platform. This could be used to detect malnutrition in children under five years and provide an easy way to transfer nutrition advice to citizens through the e-government initiatives. Furthermore the proposed system illustrates that by increasing the number of rules to make the final decision it will increase the number of records which get the correct final decision. For this purpose the rules will be generated by using the sequential covering algorithm. Furthermore rules will be categorized based on the classes which are associating with the rules. The final decision will be based on number of rules which is generated by the system. The rules which use to make the final decision are based on the likelihood ratio for each rule. The same process will be continuing by changing the number of rules which is used to make the final decision.Mainly the testing is done only on the rule comparison part of the system. The entire training set consists of 140 records. Among that 71 records are belong to malnutrition class and 69 records are belong to the nutrition class. Thereafter rules are sorted according to the descending order of the likelihood ratio and best 30 rules are used to make the final decision of the test set. Test set consist of 50 records. Likewise same procedure is carried out for 35 rules,40 rules,45 rules and 50 rules. The results shows that the numbers of records getting the correct answer will increase if we increase the number of rules used to make the final decision.The goal of this research is to contribute in the reduction of malnutrition in Sri Lanka to some extent by improving the transfer of nutrition knowledge among the citizens. It also intends to provide an effective method of finding whether a child is having malnutrition or not in an easy pace.