Association Rules

Multi-agent systems underpin the vision for ambient intelligence. However, developing multi-agent systems is a complex and challenging process. For example, pervasive computing has been found susceptible to instability, due to unwanted behaviour arising from unplanned interaction between rule based agents. This instability is impossible to predict, as it depends on the rules of interaction, the initial state of the system, the user interaction, and in the time delay of the system (due to network traffic, different speed of processing, etc). In this paper we present a theoretical framework, an Interaction Network (IN), together with a communication locking strategy that we call INPRES (Instability Prevention System) that can be used to identify and eliminate this problem. In addition we describe a Multi-Dimensional Model (MDM) to represent the agents and the state of each agent over time. A theorem showing the role of delays in an unstable system is presented. We present experimental results based on simulations and a physical emulation that demonstrate the effectiveness of these methods.

Association rules discovered from transaction databases can be large in number. Reduction of association rules is an issue in recent times. Conventionally by varying support and confidence number of rules can be increased and decreased. By combining additional constraint with support number of frequent itemsets can be reduced and it leads to generation of less number of rules. Average inter itemset distance(IID) or Spread, which is the intervening separation of itemsets in the transactions has been used as a measure of interestingness for association rules with a view to reduce the number of association rules. In this paper by using average Inter Itemset Distance a complete algorithm based on the apriori is designed and implemented with a view to reduce the number of frequent itemsets and the association rules and also to find the distribution pattern of the association rules in terms of the number of transactions of non occurrences of the frequent itemsets. Further the apriori algorithm is also implemented and results are compared. The theoretical concepts related to inter itemset distance are also put forward.

Educational Institutions tend to find ways to improve academic performance by implementing different analytical tools and techniques. The explosive size of educational databases has many advantages and one of them is they can be used to enhance the performance of all academic staff and the institutions in the result. Educational Data Mining (EDM) as a new trend of data mining has faced a huge number of researches and tested almost all data mining techniques and algorithms which result in emerging enhancements in all educational fields. One of the most important functions in data mining is association rules mining. The association rules mining finds the association between dataset features and produces the association as rules. The resulting rules describe many hidden patterns inside the dataset and this feature is very helpful in EDM. Since all educational databases have many features which make the resulting association rules too many, so there is a need for using Feature Selection (FS) algorithm to reduce the features and use important features only. FS has become increasingly important as the size and dimensionality of educational datasets increase. FS is one of most important data mining research areas. It aims to identify several features that describe the dataset better than the original set of features. This objective can be achieved by removing redundant and less correlated features according to importance criteria in FS. Various selection algorithms have been proposed for filter-based FS. ReliefF is one of the most important algorithms that have been successfully implemented in many FS applications. This study presents a new model that can be utilized by any university that seeks to improve the quality of education by analyzing data and identifying factors that affect academic results to increase students' chances of success. We focus on Relief-based algorithms (RBAs) according to the technique of mining association rules. The main concept of the proposed algorithm is to find features that are closely correlative with the class attribute through the mining of association rules. Experimental results based on several real datasets confirm that the proposed model can obtain a smaller and promising feature subset compared with other models and techniques. The outcomes of this research can help improve policies in higher education.

Association Rule discovery has been an important problem of investigation in knowledge discovery and data mining. An association rule describes associations among the sets of items which occur together in transactions of databases.The Association Rule mining task consists of finding the frequent itemsets and the rules in the form of conditional implications with respect to some prespecified threshold values of support and confidence.The interestingness of Association Rules are determined by these two measures. However, other measures of interestingness like lift and conviction are also used. But, there occurs an explosive growth of discovered association rules and many of such rules are insignificant. In this paper we introduce a new measure of interestingness called Inter Itemset Distance or Spread and implemented this notion based on the approaches of the apriori algorithm with a view to reduce the number of discovered Association Rules in a meaningful manner. An analysis of the working of the new algorithm is done and the results are presented and compared with the results of conventional apriori algorithm.

Ambient Intelligence, and in general, any autonomous rule based system has been found to suffer from cyclic instability. This behaviour is characterized by unwanted oscillations, due to interacting rules within networks of pervasive computing devices. The binary behaviour of each agent is defined via a set of boolean rules, and the behaviour of the system as a whole is given by the ensemble of rules defined over the set of agents. From complex theory it has been found that the problem of cyclic instability cannot be solved analytically; however, it is possible to prevent it. In this paper we present a novel solution based on locking, to prevent cyclic instability. This strategy makes use of the topological properties of the digraph associated called Interaction Network (IN), and the local rules of the interacting agents. The concept of strong and weak coupling is introduced. Using the strong and weak concepts, a strategy c-INPRES that minimizes the number of agents locked is presented. Preliminary and encouraging results are shown.