Systematic Construction of Goal-Oriented COTS Taxonomies

Informe de investigación realizado a partir de una estancia en el Department of Computer and Information Science de la Norwegian University of Science and Technology (NTNU), Noruega, entre setiembre i diciembre de 2006. El uso de componentes de software llamados Commercial-Off-The-Shelf (COTS) en el desarrollo de sistemas basados en componentes implica varios retos. Uno de ellos es la falta de información disponible y adecuada para dar soporte al proceso de selección de componentes a ser integrados. Para lidiar con estos problemas, se esta desarrollando un trabajo de tesis que propone un método llamado GOThIC (Goal-Oriented Taxonomy and reuse Infrastructure Construction). El método está orientado a construir una infrastructura de reuse para facilitar la búsqueda y reuse de componentes COTS. La estancia en la NTNU, reportada en este documento, tuvo como objetivo primordial las mejora del método y la obtención de datos empíricos para darle soporte. Algunos de los principales resultados fueron la obtención de datos empíricos fundamentando la utilización del método en ámbitos industriales de selección de componentes COTS, así como una nueva estrategia para conseguir de forma factible e incremental, la federación y reuso de los diferentes esfuerzos existentes para encontrar, seleccionar y mantener componentes COTS y Open Source (OSS) -comúnmente llamados componentes Off-The-Shelf (OTS) - en forma estructurada.

Combining statistical and semantic approaches to the translation of ontologies and taxonomies

Ontologies and taxonomies are widely used to organize concepts providing the basis for activities such as indexing, and as background knowledge for NLP tasks. As such, translation of these resources would prove useful to adapt these systems to new languages. However, we show that the nature of these resources is significantly different from the "free-text" paradigm used to train most statistical machine translation systems. In particular, we see significant differences in the linguistic nature of these resources and such resources have rich additional semantics. We demonstrate that as a result of these linguistic differences, standard SMT methods, in particular evaluation metrics, can produce poor performance. We then look to the task of leveraging these semantics for translation, which we approach in three ways: by adapting the translation system to the domain of the resource; by examining if semantics can help to predict the syntactic structure used in translation; and by evaluating if we can use existing translated taxonomies to disambiguate translations. We present some early results from these experiments, which shed light on the degree of success we may have with each approach

Ontologies, taxonomies, thesauri:learning from texts

The use of ontologies as representations of knowledge is widespread but their construction, until recently, has been entirely manual. We argue in this paper for the use of text corpora and automated natural language processing methods for the construction of ontologies. We delineate the challenges and present criteria for the selection of appropriate methods. We distinguish three ma jor steps in ontology building: associating terms, constructing hierarchies and labelling relations. A number of methods are presented for these purposes but we conclude that the issue of data-sparsity still is a ma jor challenge. We argue for the use of resources external tot he domain specific corpus.

Combining Taxonomies using Word2vec

Taxonomies have gained a broad usage in a variety of fields due to their extensibility, as well as their use for classification and knowledge organization. Of particular interest is the digital document management domain in which their hierarchical structure can be effectively employed in order to organize documents into content-specific categories. Common or standard taxonomies (e.g., the ACM Computing Classification System) contain concepts that are too general for conceptualizing specific knowledge domains. In this paper we introduce a novel automated approach that combines sub-trees from general taxonomies with specialized seed taxonomies by using specific Natural Language Processing techniques. We provide an extensible and generalizable model for combining taxonomies in the practical context of two very large European research projects. Because the manual combination of taxonomies by domain experts is a highly time consuming task, our model measures the semantic relatedness between concept labels in CBOW or skip-gram Word2vec vector spaces. A preliminary quantitative evaluation of the resulting taxonomies is performed after applying a greedy algorithm with incremental thresholds used for matching and combining topic labels.

A contribution to a virtual enterprise taxonomy

This paper presents a taxonomy able to contribute to building a framework within the domain of Virtual Enterprises (VE). A VE taxonomy currently does not exist, and this lack is felt in the ambiguous way that some concepts are addressed, leading to a fragment understanding that hinders the development of the science of VE integration and management. The structure of the taxonomy developed is based on the view of the system as a 5-tuple consisting of Input, Control, Output, Mechanism, and Process, which is the underlying system-view in the well-know IDEF0 diagramming technique. In particular, this taxonomy addresses the VE extended lifecycle that implies the use of a meta-organization called Market of Resources, as an original contribution to the VE theory and practice. The taxonomy presented is constructed in a way to be easily complemented with other VE partial taxonomies that may be found in literature.

A taxonomy for virtual enterprises

The purpose of this paper is to present a taxonomy able to contribute to building a framework within the domain of Virtual Enterprises (VE), to facilitate the sharing of knowledge and contributions to knowledge, as well as for trust building among VE stakeholders. A VE taxonomy currently does not exist, and this lack is felt in the ambiguous way that some concepts are addressed, leading to a fragment understanding that hinders the development of the science of VE integration and management. The structure of the taxonomy developed is based on the view of the system as a 5-tuple consisting of Input, Control, Output, Mechanism, and Process, which is the underlying system-view in the well-know IDEF0 diagramming technique. In particular, this taxonomy addresses the VE extended lifecycle that implies the use of a meta-organization called Market of Resources, as an original contribution to the VE theory and practice. The taxonomy presented does not repeat what the literature already includes, or the commonplaces, and it is constructed in a way to be easily complemented with other VE partial taxonomies that may be found in literature. Some suggestions for extensions to other interrelated domains (as evolution leaves taxonomies in an open or incompleteness state) are given in the text.

Transacção de componentes multimédia suportada por agentes

O objectivo do projecto descrito nesta dissertação é o desenvolvimento da interface entre as empresas e a plataforma Business-to-Business (B2B) de negociação automática de anúncios em construção. A plataforma, no seu todo, deve garantir que os intervalos da programação são preenchidos com um alinhamento de anúncios compatível com os interesses expressos e o perfil construído dos espectadores. A plataforma funciona como um mercado electrónico de negociação automática destinado a agências de publicidade (empresas produtoras) e empresas provedoras de conteúdos e serviços multimédia aos consumidores finais (empresas distribuidoras). As empresas, uma vez registadas na plataforma, passam a ser representadas por agentes que negoceiam automaticamente os itens submetidos com o comportamento especificado. Do ponto de vista da arquitectura, a plataforma consiste num sistema multiagente organizado em três camadas compostas por: (i) agentes de interface com as empresas; (ii) agentes de modelação das empresas; e (iii) agentes delegados, de duração efémera, exclusivamente criados para participar em negociações específicas de conteúdos multimédia. Cada empresa representada na plataforma possui, para além de um número indeterminado de delegados envolvidos em negociações específicas, dois agentes: (i) o agente de interface com a empresa, que expõe um conjunto de operações de interface ao exterior através de um serviço Web, localizado na primeira camada; e (ii) o agente que modela a empresa na plataforma, que expõe através de um serviço Web um conjunto de operações aos agentes das restantes camadas da plataforma, residente na camada intermédia. Este projecto focou-se no desenvolvimento da camada superior de interface da plataforma com as empresas e no enriquecimento da camada intermédia. A realização da camada superior incluiu a especificação da parte da ontologia da plataforma que dá suporte às operações de interface com o exterior, à sua exposição como serviços Web e à criação e controlo dos agentes de interface. Esta camada superior deve permitir às empresas carregar e descarregar toda informação relevante de e para a plataforma, através de uma interface gráfica ou de forma automática, e apresentar de forma gráfica e intuitiva os resultados alcançados, nomeadamente, através da apresentação da evolução das transacções. Em relação à camada intermédia, adicionou-se à ontologia da plataforma a representação do conhecimento de suporte às operações de interface com a camada superior, adoptaram-se taxonomias de classificação de espectadores, anúncios e programas, desenvolveu-se um algoritmo de emparelhamento entre os espectadores, programas e anúncios disponíveis e, por fim, procedeu-se ao armazenamento persistente dos resultados das negociações. Do ponto de vista da plataforma, testou-se o seu funcionamento numa única plataforma física e assegurou-se a segurança e privacidade da comunicação entre empresa e plataforma e entre agentes que representam uma mesma empresa.

Persisting problems related to race and ethnicity in public health and epidemiology research

A recent and comprehensive review of the use of race and ethnicity in research that address health disparities in epidemiology and public health is provided. First it is described the theoretical basis upon which race and ethnicity differ drawing from previous work in anthropology, social science and public health. Second, it is presented a review of 280 articles published in high impacts factor journals in regards to public health and epidemiology from 2009-2011. An analytical grid enabled the examination of conceptual, theoretical and methodological questions related to the use of both concepts. The majority of articles reviewed were grounded in a theoretical framework and provided interpretations from various models. However, key problems identified include a) a failure from researchers to differentiate between the concepts of race and ethnicity; b) an inappropriate use of racial categories to ascribe ethnicity; c) a lack of transparency in the methods used to assess both concepts; and d) failure to address limits associated with the construction of racial or ethnic taxonomies and their use. In conclusion, future studies examining health disparities should clearly establish the distinction between race and ethnicity, develop theoretically driven research and address specific questions about the relationships between race, ethnicity and health. One argue that one way to think about ethnicity, race and health is to dichotomize research into two sets of questions about the relationship between human diversity and health.

Construção de um manual de codificação e interpretação de memórias autobiográficas

Dissertação de mestrado integrado em Psicologia

O professor como construtor do currículo: integração da tecnologia em atividades de aprendizagem de matemática

O desenvolvimento profissional dos professores de matemática, por meio de progra¬mas nacionais e formações contínuas, deve proporcionar experiências que envolvam investigação, pensamento, planeamento, prática e reflexão. No caso da tecnologia, não nos devemos focar nas ferramentas em si, mas no modo como são usadas pelos docentes em contexto de sala de aula. Existem taxonomias de atividades de apren¬dizagem baseadas no conteúdo assentes na ideia do professor como construtor do currículo, que, para integrar com sucesso a tecnologia educativa nas aulas, desenvolve o conhecimento pedagógico e tecnológico do conteúdo (TPACK), e apresenta-se a de matemática. Desse modo, reflete-se, por meio de vários estudos nacionais e internacionais, que as tecnologias deverão ser usadas pelos professores de acordo com objetivos, conteúdos e pedagogias específicas para terem um efeito positivo na aprendizagem dos alunos sobre as atividades baseadas no conteúdo que melhor se enquadram com essas tecnologias.

Rate of gene sequence evolution and species diversification in flowering plants: a re-evaluation

Barraclough and co-workers (in a paper published in 1996) observed that there was a significant positive correlation between the rate of evolution of the rbcL chloroplast gene within families of flowering plants and the number of species in those families. We tested three additional data sets of our own (based on both plastid and nuclear genes) and used methods designed specifically for the comparison of sister families (based on random speciation and extinction). We show that, over all sister groups, the correlation between the rate of gene evolution and an increased diversity is not always present. Despite tending towards a positive association, the observation of individual probabilities presents a U-shaped distribution of association (i.e. it can be either significantly positive or negative). We discuss the influence of both phylogenetic sampling and applied taxonomies on the results.

Analyzing hidden semantics in social bookmarking of open educational resources

Web 2.0 services such as social bookmarking allow users to manage and share the links they find interesting, adding their own tags for describingthem. This is especially interesting in the field of open educational resources, asdelicious is a simple way to bridge the institutional point of view (i.e. learningobject repositories) with the individual one (i.e. personal collections), thuspromoting the discovering and sharing of such resources by other users. In this paper we propose a methodology for analyzing such tags in order to discover hidden semantics (i.e. taxonomies and vocabularies) that can be used toimprove descriptions of learning objects and make learning object repositories more visible and discoverable. We propose the use of a simple statistical analysis tool such as principal component analysis to discover which tags createclusters that can be semantically interpreted. We will compare the obtained results with a collection of resources related to open educational resources, in order to better understand the real needs of people searching for open educational resources.