Label matching

Schema matching is a basic problem of database application, semantic web, web services in e-business, semantic-based query and etc. Today's, matching implementation is mostly done in manually. The other side, some research have designed some technique for supporting the schema matching automation in some domain. Some approaches [4], [5] have been existed for conducting the schema matching such as schema level, instance level, language based approach and so on. In this paper, we will test and investigate language based and string based method for the source which written in Indonesia language. We will do schema matching based on terminology approach for testing sources of Indonesia language case. The reason is that most of the approach is English based. The Approach process will utilize on-line dictionary which combine with WordNet with path length method as input in language based. Matching result of our approach will be evaluated by comparing with manually result matching. Our evaluation can give the contribution to be developed as model for mapping in interoperability on Indonesia language information source which more dynamic, semi structure and opened in Internet, as well as semantic web application or web services.

E↵ective matching of activities is the first step toward successful process model matching and search. The problem is nontrivial and has led to a variety of computational similarity metrics and matching approaches, however all still with low performance in terms of precision and recall. In this paper, instead, we study how to leverage on human intelligence to identify matches among activities and show that the problem is not as straightforward as most computational approaches assume. We access human intelligence (i) by crowdsourcing the activity matching problem to generic workers and (ii) by eliciting ground truth matches from experts. The precision and recall we achieve and the qualitative analysis of the results testify huge potential for a human-based activity matching that contemplates disagreement and interpretation.

In this paper we present a novel algorithm, named GBAP, that jointly uses automatic programming with ant colony optimization for mining classification rules. GBAP is based on a context-free grammar that properly guides the search process of valid rules. Furthermore, its most important characteristics are also discussed, such as the use of two different heuristic measures for every transition rule, as well as the way it evaluates the mined rules. These features enhance the final rule compilation from the output classifier. Finally, the experiments over 17 diverse data sets prove that the accuracy values obtained by GBAP are pretty competitive and even better than those resulting from the top Ant-Miner algorithm.

E↵ective matching of activities is the first step toward successful process model matching and search. The problem is nontrivial and has led to a variety of computational similarity metrics and matching approaches, however all still with low performance in terms of precision and recall. In this paper, instead, we study how to leverage on human intelligence to identify matches among activities and show that the problem is not as straightforward as most computational approaches assume. We access human intelligence (i) by crowdsourcing the activity matching problem to generic workers and (ii) by eliciting ground truth matches from experts. The precision and recall we achieve and the qualitative analysis of the results testify huge potential for a human-based activity matching that contemplates disagreement and interpretation.

Establishment of institutions of higher learning requires massive amount of different resources which are always in short supply. The delivery of learning material and tests to the students has become very easy with the facility of uploading the same on the web irrespective of the number of students. The assessment part could be a deterrent as far as willingness of learned faculty members to participate in the whole process is concerned. If assessment will become automated then it will be easier for any teachers to evaluate any number of students. This paper presents a proposed architecture of automated assessment of Use-Case Diagram. The essence of this architecture is to assess large number of students very easily in short duration. This proposed work is going to be very useful for the needy students by assisting in evaluation of their performance.

In this paper we present a novel algorithm, named GBAP, that jointly uses automatic programming with ant colony optimization for mining classification rules. GBAP is based on a context-free grammar that properly guides the search process of valid rules. Furthermore, its most important characteristics are also discussed, such as the use of two different heuristic measures for every transition rule, as well as the way it evaluates the mined rules. These features enhance the final rule compilation from the output classifier. Finally, the experiments over 17 diverse data sets prove that the accuracy values obtained by GBAP are pretty competitive and even better than those resulting from the top Ant-Miner algorithm.

Patrouiller dans un environnement implique une équipe d'agents dont le but consiste à visiter continuellement et aussi fréquemment que possible les lieux les plus pertinents. Afin d'obtenir des performances optimales, il est alors primordial que les agents coordonnent leurs actions. De nombreux domaines peuvent être concernés par ce problème, comme la robotique, la simulation ou les jeux vidéo. Nous adoptons dans cet article une approche d'optimisation basée sur les colonies de fourmis pour traiter ce problème. Deux algorithmes sont proposés, dans lesquels des colonies de fourmis sont engagées dans une compétition pour découvrir la meilleure stratégie de patrouille multi-agent. Les résultats expérimentaux montrent que, sur quatre des six graphes étudiés, l'une de nos techniques est significativement meilleure que la technique d'apprentissage par renforcement proposée par Santana. ABSTRACT. Patrolling an environment involves a team of agents whose goal usually consists of continuously visiting its most relevant areas as frequently as possible. For such a task, agents have to coordinate their actions in order to achieve optimal performance. A wide range of applications can be dealt with this problem, from computer network management to vehicle routing. The Ant Colony Optimization is adopted here as the solution approach to this problem. Two novel ACO algorithms are proposed here, in which several ants' colonies try to discover the best multi-agent patrolling strategy. Experimental results show that, for four out of the six evaluated graphs, one of our techniques significantly outperforms the reinforcement learning technique proposed by Santana, irrespective of the number of the involved patrolling agents. MOTS-CLÉS : patrouille multi-agent, colonies de fourmis, ACO.

In this paper we present a novel algorithm, named GBAP, that jointly uses automatic programming with ant colony optimization for mining classification rules. GBAP is based on a context-free grammar that properly guides the search process of valid rules. Furthermore, its most important characteristics are also discussed, such as the use of two different heuristic measures for every transition rule, as well as the way it evaluates the mined rules. These features enhance the final rule compilation from the output classifier. Finally, the experiments over 17 diverse data sets prove that the accuracy values obtained by GBAP are pretty competitive and even better than those resulting from the top Ant-Miner algorithm.

E↵ective matching of activities is the first step toward successful process model matching and search. The problem is nontrivial and has led to a variety of computational similarity metrics and matching approaches, however all still with low performance in terms of precision and recall. In this paper, instead, we study how to leverage on human intelligence to identify matches among activities and show that the problem is not as straightforward as most computational approaches assume. We access human intelligence (i) by crowdsourcing the activity matching problem to generic workers and (ii) by eliciting ground truth matches from experts. The precision and recall we achieve and the qualitative analysis of the results testify huge potential for a human-based activity matching that contemplates disagreement and interpretation.

Reconciliation of billing and payments is a critical process in the processing of medical reimbursement data for which no satisfactory solution exists so far. Such reconciliation aims at alleviating the manual work for operator, and is characterized by a high expectation with regard to the algorithm effectiveness: automated reconciliation can only be proposed if it is almost certain, so as to avoid costly and un-ergonomic undo process. A promissing solution for performing automated reconciliation is the use of Ant-Colony Optimisation (ACO)-based learning, which has a strong ability of finding sequences of evaluation criteria, here in the context of label matching, while taking into account search history weighted by success scores and history. The ant reconciliation algorithm (ARA) extends and implements the Ant-Miner algorithm by defining specific computation parameters as well as a rule model for text matching. We define the ARA model, explicit its implementation, and provide evidences of the relevance of the proposal drawn from evaluation on real world medical billing and payment data.

Une des difficultés inhérentes à la recherche énumérative est l'explosion combinatoire. Parmi les algorithmes incomplets qui tentent de résoudre ce problème, l'optimisation par colonie de fourmis (ACO -Ant Colony Optmisation), qui combine des méthodes aléatoires et heuristiques avec l'apprentissage par renforcement, a prouvé son efficacité sur de nombreux problèmes de satisfaction de contraintes (CSP). Cet article présente une application d'un algorithme basé sur ACO pour la configuration, ce qui à notre connaissance n'avait pas encore été étudié. Nous décrivons comment la nature des problèmes non-bornés de configuration influe sur l'approche ACO, notamment à cause de la présence de variables ensemblistes et de domaines ouverts. Nous proposons un algorithme et un modèle phéromonal original permettant de traiter ces difficultés. Nous montrons également l'utilisation de l'optimisation par essaim de particules (PSO) pour converger vers des ensembles de paramètres optimaux. Enfin, nous fournissons des résultats expérimentaux, à la fois pour des instances aléatoires et pour le problème d'optimisation des racks.

Les algorithmes basés sur l'optimisation par colonies de fourmis (Ant Colony Optimization / ACO) sont paramétrés par deux coefficients α et β qui déterminent respectivement l'importance de l'expérience passée et de l'heuristique dans la probabilité de transition utilisée pour construire des solutions. Nous introduisons une méthode pour adapter dynamiquement les valeurs de ces deux paramètres qui ont une influence déterminante sur l'intensification et la diversification de la recherche. Cette méthode est ellemême basée sur ACO : des traces de phéromone sont associées aux valeurs potentielles des paramètres ; ces traces représentent l'expérience passée concernant l'utilisation de ces valeurs et sont utilisées pour adapter dynamiquement α et β. Nous proposons deux variantes qui diffèrent par la granularité de l'apprentissage : dans un cas l'apprentissage est global à l'ensemble des variables du problème, dans l'autre, il est spécifique à chaque variable. Nous évaluons et comparons expérimentalement ces deux méthodes dans le cadre d'une mise en oeuvre de la méthode ACO pour résoudre des problèmes de satisfaction de contraintes.