K. Revoredo

ABSTRACT The regression-by-discretization approach allows the use of classification algorithm in a regression task. It works as a pre-processing step in which the numeric target value is discretized into a set of intervals. We had applied this approach to the Hidden Markov Model for Regression (HMMR) which was successfully compared to the Naive Bayes for Regression and two traditional forecasting methods, Box-Jenkins and Winters. In this work, to further improve these results, we apply three discretization methods to HMMR using ten time series data sets. The experimental results showed that one of the discretization methods improved the results in most of the data sets, although each method improved the results in at least one data set. Therefore, it would be better to have a search algorithm to automatically find the optimal number and width of the intervals.

ABSTRACT A business process is a set of activities performed in a coordinated manner within an organizational and technical environment that is aimed toward a business goal. The flexibility of a process is related to an understanding of the unexpected events that occur when people, systems and resources interact and require adjustments. Thus, business processes must be designed to respond to information about different events and their specificity. This information defines what the literature calls “context”. To broaden the perception of context in the case of a business process, this work proposes an approach to characterize the context of a business process activity in a given domain through conceptual models structured in layers. A case study was conducted to evaluate the proposal, which provided evidence of the applicability of the model.

ABSTRACT Dynamic adaptation is the customization of a business process to make it applicable to a particular situation at any time of its life cycle. Adapting requires experience, and involves knowledge about various, internal and external, aspects of business. Thus, we argue for the application of adaptation rules, considering the context of a particular process instance. Furthermore, we state that a context-based adaptation environment should go beyond, and learn from decisions, as well as continuously identify new unforeseen situations (context definitions). The aim of this paper is to present a computational engine that infers the need to update situations and adaptation rules, suggesting changes to them. An application scenario is presented to discuss the usage of the proposal.

ABSTRACT This paper presents the ontology-related research themes being conducted at NP2Tec/UNIRIO. The main focuses that are being investigated handle the ontological representation of Enterprise artifacts, including structural conceptual models, process models, and business rules. Besides, maintaining a consistent representation of a Universe of Discourse is also a challenge since new (valid) facts are constantly being observed, and may not be consistent with the existing ontological representation.

... 1 Fernando Rebouças Stucchi Flávio Buiochi ... Pedro Pereira de Paula 1 Sérgio Luís Lopes Verardi 1 Yasmara Conceicao De Polli Migliano 1 Antonio Carlos de Jesus Paes 1 Djonny Weinzierl 1 Fabio Henrique Pereira 1 Mauricio Barbosa de Camargo Salles 1 ...

ABSTRACT Ontology alignment is the process of finding corresponding entities with the same intended meaning in different ontologies. In scenarios where an ontology conceptually describes the contents of a data repository, this provides valuable information for the purpose of semantic data integration which, in turn, is a fundamental issue for improving business intelligence. A basic, yet unsolved, issue in semantic data integration is how to ensure that only data related to the same real-world entity is merged. This requires that each concept is precisely defined into the ontology. From another perspective, foundational ontologies describe very general concepts independent of a particular domain, and precisely define concept meta-properties so as to make the semantics of each concept in the ontology explicit. In this paper, we discuss how the use of foundational ontology can improve the precision of the ontology alignment, and illustrate some examples using the UFO foundational ontology.

... Page 14. 236 Kate Revoredo and Gerson Zaverucha Acknowledgements ... 27. D. Poole. Learning, Bayesian Probability, Graphical Models, and Abduction. P. Flach and A. Kakas, editors, Abduction and Induction: essays on their relation and integration. Kluwer, 1998. 28. ...

ABSTRACT Foundational Ontologies help maintaining and expanding ontologies expressivity power, thus enabling them to be more precise and free of ambiguities. The use of modeling languages based on these ontologies, such as OntoUML, requires not only the modeler's experience regarding such languages, but also a good understanding about the domain being modeled. Aiming to facilitate, or even enable the modeling of complex domains, several techniques have been proposed in order to automatically generate ontologies from texts. However, none is able to generate well-founded ontologies (which are constructed based on Foundational Ontologies). Moreover, an important issue on learning from text is how to distinguish among different meanings of a word, which impacts on concepts expressed by the ontologies. Therefore, techniques for word sense disambiguation must be considered. This paper proposes a technique for automatically learn well-founded ontologies described in OntoUML through word sense disambiguation.

ABSTRACT Description Logics based languages have emerged as the standard knowledge representation scheme for ontologies. Typically, an ontology formalizes a number of dependent and related concepts in a domain, encompassed as a terminology. As defining such terminologies manually is a complex, time consuming and error-prone task, there is great interest and even demands for methods that learn terminologies automatically. Learning a terminology in Descriptions Logics concerns to learn several related concepts. This process would greatly benefit of an ideal order to determine which concept should be learned before another concept. Arguably, such an order would yield rich and readable terminologies, as previously, and interrelated concepts formerly learned could be used to induce the description of further concepts. In this work, we contribute with a formal definition of the concept and terminology learning problems and from such definitions we devise an algorithm for finding an ordering through concept taxonomy discovery, that should be followed when learning several related concepts. We show through an experiment that by following the order detected by the algorithm, we are able to afford a more readable terminology than methods that do not conceive an ideal order or do not learn concepts in a dependent way.

Resumo A arquitetura orientada a serviços (SOA–Service Oriented Architecture) apresenta-se como sendo mais flexível e capaz de suportar serviços independentes de plataforma e protocolo em um ambiente distribuído. Neste trabalho analisamos modelos de ciclos de ...

Abstract. The complexity and richness of geospatial data and the difficulty of their representation create specific problems for geographic information systems interoperability. This paper proposes the creation of software components from ontologies as a way to integrate geographic information. These software components are derived from ontologies using an object-oriented mapping. The translation of ontologies into active information system components leads to ontology-driven geographic information systems. The ...

Abstract. There has been a lot of work in the integration of probabilistic reasoning with first order logic representations. Learning algorithms for these models have been developed and they all considered modifications in the entire structure. Previously Revoredo and Zaverucha argued that when the theory is approximately correct the use of techniques from theory revision to just modify the structure in places that failed in logically covering the examples can be a more adequate choice. In the present paper, we extend this revision system ...

Resumo. Durante o processo de revisao, similarmente quando aprendendo em ILP, o algoritmo pode nao ser capaz de propor uma modifica�� ao ��til usando a linguagem existente. Neste caso, t��cnicas de inven�� ao de predicados podem ser usadas para automaticamente estender a linguagem com a defini�� ao de novos predicados. Neste trabalho, estendemos o nosso sistema de revisao de teorias probabil��sticas de primeira-ordem, chamado PFORTE, com dois novos operadores de revisao que propoem novos ...

Resumo. Uma questao importante em aprendizado de redes Bayesianas (RB) �� como aprender a estrutura da rede na presen��a de vari��veis nao-observadas. Neste artigo, propomos uma nova abordagem que utiliza t��cnicas de revisao de teoria. O algoritmo proposto, denominado DAHVI, aplica uma heur��stica discriminativa ea partir dos exemplos identifica pontos potenciais na RBa inclusao de uma vari��vel nao-observada. Essas vari��veis sao inclu��das atrav��s de um operador de revisao proposto neste artigo. O ...