Despite the many clear advantages that ontology has to offer as a standardized knowledge representation language, many intelligent system developers still find it difficult to jump on the band wagon and represent all their application knowledge using ontology. This difficulty and hesitation stems primarily from the fact that, while most ontology languages provide native support for reasoning about the domain's structures, they do not provide adequate support for computational planning -- the kind of reasoning used by many intelligent systems to derive their purposeful behaviors. To overcome this challenge, a lot of work has been done to discover a practical way to seamlessly incorporate planning into ontology languages. As it has been well-established in the literature however, this is a very challenging task from both a theoretical and practical stand point, and many of the reported works in this direction either have had very limited success, or have been done in ad hoc and less reusable manners. In this thesis, we report our pursuit of a new approach to integrating planning into ontology-driven applications. This approach promises to overcome the difficulties faced by many of the existing approaches. In addition to producing a reusable and extensible framework for doing computational planning in ontology-driven applications, our pursuit also raises and answers some interesting ontology research questions that could have potential impacts on several application domains beyond the integration of planning and ontological modeling.

<p>  </p> <p>The abundance of heterogeneous data produced and collected each day via multimodal sources may contain hidden events of interest, but in order to extract them the streams of data need to be analyzed with appropriate algorithms, so these events are presented to the end user at the right moment and at the right time. This dissertation proposes a series of algorithms that shape a comprehensive framework for situational knowledge on demand to address this problem. The framework consists of several modules and approaches, each of them is presented in a separate chapter: I begin with video data analysis in streaming video and video at rest for enhanced object detection of real-life surveillance video. For detecting the rare events of interest, I develop a semantic video analysis algorithm which uses an overlay knowledge graph and a semantical network. I show that the usage of the external knowledge for understanding the semantic analysis outperforms other techniques such as transfer learning. </p> <p>The semantical outliers can be used further for improving the algorithm of detecting new objects in the stream of different modalities. I extend the framework with additional modules for natural language data and apply the extended version of the semantic analysis algorithm to define the events of interest from multimodal streaming data. I present a way of combining several feature extractors which can be extended to multiple heterogeneous streams of data in order to efficiently fuse the data based on its semantical similarity, and then show how the serverless architecture of the framework outperforms conventional cloud software architecture. </p> <p>Besides detecting the rare and semantically incompatible events, the semantic analysis can be used for improving the neural networks performance with the data augmentation. The algorithm presented for augmenting the data with the potentially novel objects to circumvent the data drift problem uses the knowledge graph and generative adversarial networks to present the objects to augment the training datasets for supervised learning. I extend the presented framework with a pipeline for generating synthetic novelties to improve the performance of feature extractors and provide the empirical evaluation of the developed method.</p>

La problématique d'interrogation d'une base de données en présence d'une ontologie (OBQA pour "Ontology-based Query Answering") consiste à prendre en compte des connaissances générales, typiquement une ontologie de domaine, lors de l'évaluation d'une requête. Dans le cadre de cette thèse, ces connaissances sont représentées par des formules de la logique du premier ordre appelées "règles existentielles". Les règles existentielles sont aussi connues sous le nom de règles Datalog+/- ou "tuple-generating dependencies". Nous considérons une approche couramment utilisée, qui consiste à réécrire la requête en exploitant les règles de façon à se ramener à un problème classique l'interrogation d'une base de données. Nous définissons un cadre théorique d'étude des algorithmes de réécriture d'une requête conjonctive en une union de requêtes conjonctives, accompagné d'un algorithme de réécriture générique, prenant en paramètre un opérateur de réécriture. Nous proposons ensuite plusieurs opérateurs de réécriture et développons différentes optimisations, que nous évaluons sur des benchmarks du domaine. / The issue of querying a knowledge base, also called Ontology-based Query Answering (OBQA), consists of taking into account general knowledge, typically a domain ontology, when evaluating queries. In this thesis, ontological knowledge is represented by first-order logical formulas, called existential rules. Existential rules are also known as Datalog+/- and tuple-generating dependencies. We adopt a well-known approach, which consists of rewriting the query with the rules to reduce the problem to a classical database query answering problem. We define a theoretical framework to study algorithms that rewrite a conjunctive query into a union of conjunctive queries, as well as a generic rewriting algorithm that takes into account a rewriting operator. Then, we propose several rewriting operators and develop several optimisations, which we evaluate on benchmarks of the domain.

Seres humanos são capazes de desenvolver complexas estruturas de conhecimento que podem ser utilizadas de modo flexível para lidar com o ambiente de maneira apropriada. Estas estruturas de conhecimento constituem um núcleo que suporta processos cognitivos, tais como a percepção, a categorização, o planejamento, etc. A Inteligência Artificial, enquanto área de investigação, ocupa-se de desenvolver meios que viabilizem a reprodução destas capacidades cognitivas em agentes artificiais. Por este motivo, a investigação de abordagens que permitam a representação de conhecimento de um modo flexível se revela altamente relevante. Com o objetivo de superar algumas das limitações típicas da teoria clássica, que é adotada por várias abordagens propostas na Inteligência Artificial, este trabalho propõe um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio que integra aspectos de três diferentes teorias cognitivas a respeito de como conceitos são representados na cognição humana: teoria clássica, teoria do protótipo e teoria do exemplar. O arcabouço resultante é capaz de suportar a composicionalidade, a tipicalidade, a representação de instâncias atípicas dos conceitos, e a representação da variabilidade de indivíduos classificados por cada conceito. Consequentemente, o arcabouço proposto também suporta raciocínio lógico e baseado em similaridade. As principais contribuições deste trabalho são a concepção teórica e a formalização de um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio. Uma outra contribuição deste trabalho é uma abordagem de raciocínio para classificação que utiliza a abordagem de representação de conhecimento proposta. Além disso, este trabalho também apresenta duas abordagens para seleção de exemplares representativos de cada conceito e uma abordagem para extração de protótipos de conceitos. Nesta tese também é apresentado um sistema para interpretação automática de processos deposicionais que adota o arcabouço proposto. Experimentos realizados em uma tarefa de classificação sugerem que o arcabouço proposto é capaz de oferecer classificações mais informativas que as oferecidas por uma abordagem puramente clássica. / Human beings can develop complex knowledge structures that can be used for dealing with the environment in suitable ways. These knowledge structures constitute a core that supports several cognitive processes, such as perception, categorization, planning, etc. The Artificial Intelligence, as a research field, aims at developing approaches for mimicking these cognitive capabilities in machines. Due to this, it is important to investigate approaches that allow representing the knowledge in flexible ways. In order to overcome some limitations of the classical theory of knowledge representation, which is adopted by several approaches proposed in the Artificial Intelligence field, this work proposes a cognitively-inspired framework for knowledge representation and reasoning which integrates aspects from three different cognitive theories about concept representation in the human cognition: classical theory, prototype theory and exemplar theory. The resulting framework can support compositionality, typicality, representation of atypical instances of concepts, and representation of the variability of the individuals classified by each concept. Consequently, the proposed framework also supports logical reasoning and similarity-based reasoning. The main contributions of this work are the formalization of a cognitively-inspired framework for knowledge representation and reasoning, two approaches for selecting representative exemplars of each concept and an approach of reasoning for classification that integrates logical reasoning and similarity-based reasoning and that is supported by definitions, prototypes and exemplars of concepts. This thesis also presents a system for automatic interpretation of depositional processes application that adopts the proposed framework. The experiments, which were performed on a classification task, suggest that the proposed framework provides classifications that are more informative than the ones provided by a classical approach.

abstract: Different logic-based knowledge representation formalisms have different limitations either with respect to expressivity or with respect to computational efficiency. First-order logic, which is the basis of Description Logics (DLs), is not suitable for defeasible reasoning due to its monotonic nature. The nonmonotonic formalisms that extend first-order logic, such as circumscription and default logic, are expressive but lack efficient implementations. The nonmonotonic formalisms that are based on the declarative logic programming approach, such as Answer Set Programming (ASP), have efficient implementations but are not expressive enough for representing and reasoning with open domains. This dissertation uses the first-order stable model semantics, which extends both first-order logic and ASP, to relate circumscription to ASP, and to integrate DLs and ASP, thereby partially overcoming the limitations of the formalisms. By exploiting the relationship between circumscription and ASP, well-known action formalisms, such as the situation calculus, the event calculus, and Temporal Action Logics, are reformulated in ASP. The advantages of these reformulations are shown with respect to the generality of the reasoning tasks that can be handled and with respect to the computational efficiency. The integration of DLs and ASP presented in this dissertation provides a framework for integrating rules and ontologies for the semantic web. This framework enables us to perform nonmonotonic reasoning with DL knowledge bases. Observing the need to integrate action theories and ontologies, the above results are used to reformulate the problem of integrating action theories and ontologies as a problem of integrating rules and ontologies, thus enabling us to use the computational tools developed in the context of the latter for the former. / Dissertation/Thesis / Ph.D. Computer Science 2012

Seres humanos são capazes de desenvolver complexas estruturas de conhecimento que podem ser utilizadas de modo flexível para lidar com o ambiente de maneira apropriada. Estas estruturas de conhecimento constituem um núcleo que suporta processos cognitivos, tais como a percepção, a categorização, o planejamento, etc. A Inteligência Artificial, enquanto área de investigação, ocupa-se de desenvolver meios que viabilizem a reprodução destas capacidades cognitivas em agentes artificiais. Por este motivo, a investigação de abordagens que permitam a representação de conhecimento de um modo flexível se revela altamente relevante. Com o objetivo de superar algumas das limitações típicas da teoria clássica, que é adotada por várias abordagens propostas na Inteligência Artificial, este trabalho propõe um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio que integra aspectos de três diferentes teorias cognitivas a respeito de como conceitos são representados na cognição humana: teoria clássica, teoria do protótipo e teoria do exemplar. O arcabouço resultante é capaz de suportar a composicionalidade, a tipicalidade, a representação de instâncias atípicas dos conceitos, e a representação da variabilidade de indivíduos classificados por cada conceito. Consequentemente, o arcabouço proposto também suporta raciocínio lógico e baseado em similaridade. As principais contribuições deste trabalho são a concepção teórica e a formalização de um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio. Uma outra contribuição deste trabalho é uma abordagem de raciocínio para classificação que utiliza a abordagem de representação de conhecimento proposta. Além disso, este trabalho também apresenta duas abordagens para seleção de exemplares representativos de cada conceito e uma abordagem para extração de protótipos de conceitos. Nesta tese também é apresentado um sistema para interpretação automática de processos deposicionais que adota o arcabouço proposto. Experimentos realizados em uma tarefa de classificação sugerem que o arcabouço proposto é capaz de oferecer classificações mais informativas que as oferecidas por uma abordagem puramente clássica. / Human beings can develop complex knowledge structures that can be used for dealing with the environment in suitable ways. These knowledge structures constitute a core that supports several cognitive processes, such as perception, categorization, planning, etc. The Artificial Intelligence, as a research field, aims at developing approaches for mimicking these cognitive capabilities in machines. Due to this, it is important to investigate approaches that allow representing the knowledge in flexible ways. In order to overcome some limitations of the classical theory of knowledge representation, which is adopted by several approaches proposed in the Artificial Intelligence field, this work proposes a cognitively-inspired framework for knowledge representation and reasoning which integrates aspects from three different cognitive theories about concept representation in the human cognition: classical theory, prototype theory and exemplar theory. The resulting framework can support compositionality, typicality, representation of atypical instances of concepts, and representation of the variability of the individuals classified by each concept. Consequently, the proposed framework also supports logical reasoning and similarity-based reasoning. The main contributions of this work are the formalization of a cognitively-inspired framework for knowledge representation and reasoning, two approaches for selecting representative exemplars of each concept and an approach of reasoning for classification that integrates logical reasoning and similarity-based reasoning and that is supported by definitions, prototypes and exemplars of concepts. This thesis also presents a system for automatic interpretation of depositional processes application that adopts the proposed framework. The experiments, which were performed on a classification task, suggest that the proposed framework provides classifications that are more informative than the ones provided by a classical approach.

OWL 2 EL is a popular ontology language that is based on the EL family of description logics and supports regular role inclusions,axioms that can capture compositional properties of roles such as role transitivity and reflexivity. In this thesis, we present several novel complexity results and algorithms for answering expressive queries over OWL 2 EL knowledge bases (KBs) with regular role inclusions. We first focus on the complexity of conjunctive query (CQ) answering in OWL 2 EL and show that the problem is PSpace-complete in combined complexity, the complexity measured in the total size of the input. All the previously known approaches encode the regular role inclusions using finite automata that can be worst-case exponential in size, and thus are not optimal. In our PSpace procedure, we address this problem by using a novel, succinct encoding of regular role inclusions based on pushdown automata with a bounded stack. Moreover, we strengthen the known PSpace lower complexity bound and show that the problem is PSpace-hard even if we consider only the regular role inclusions as part of the input and the query is acyclic; thus, our algorithm is optimal in knowledge base complexity, the complexity measured in the size of the KB, as well as for acyclic queries. We then study graph queries for OWL 2 EL and show that answering positive, converse- free conjunctive graph queries is PSpace-complete. Thus, from a theoretical perspective, we can add navigational features to CQs over OWL 2 EL without an increase in complexity. Finally, we present a practicable algorithm for answering CQs over OWL 2 EL KBs with only transitive and reflexive composite roles. None of the previously known approaches target transitive and reflexive roles specifically, and so they all run in PSpace and do not provide a tight upper complexity bound. In contrast, our algorithm is optimal: it runs in NP in combined complexity and in PTime in KB complexity. We also show that answering CQs is NP-hard in combined complexity if the query is acyclic and the KB contains one transitive role, one reflexive role, or nominals—concepts containing precisely one individual.

Seres humanos são capazes de desenvolver complexas estruturas de conhecimento que podem ser utilizadas de modo flexível para lidar com o ambiente de maneira apropriada. Estas estruturas de conhecimento constituem um núcleo que suporta processos cognitivos, tais como a percepção, a categorização, o planejamento, etc. A Inteligência Artificial, enquanto área de investigação, ocupa-se de desenvolver meios que viabilizem a reprodução destas capacidades cognitivas em agentes artificiais. Por este motivo, a investigação de abordagens que permitam a representação de conhecimento de um modo flexível se revela altamente relevante. Com o objetivo de superar algumas das limitações típicas da teoria clássica, que é adotada por várias abordagens propostas na Inteligência Artificial, este trabalho propõe um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio que integra aspectos de três diferentes teorias cognitivas a respeito de como conceitos são representados na cognição humana: teoria clássica, teoria do protótipo e teoria do exemplar. O arcabouço resultante é capaz de suportar a composicionalidade, a tipicalidade, a representação de instâncias atípicas dos conceitos, e a representação da variabilidade de indivíduos classificados por cada conceito. Consequentemente, o arcabouço proposto também suporta raciocínio lógico e baseado em similaridade. As principais contribuições deste trabalho são a concepção teórica e a formalização de um arcabouço cognitivamente inspirado para representação de conhecimento e raciocínio. Uma outra contribuição deste trabalho é uma abordagem de raciocínio para classificação que utiliza a abordagem de representação de conhecimento proposta. Além disso, este trabalho também apresenta duas abordagens para seleção de exemplares representativos de cada conceito e uma abordagem para extração de protótipos de conceitos. Nesta tese também é apresentado um sistema para interpretação automática de processos deposicionais que adota o arcabouço proposto. Experimentos realizados em uma tarefa de classificação sugerem que o arcabouço proposto é capaz de oferecer classificações mais informativas que as oferecidas por uma abordagem puramente clássica. / Human beings can develop complex knowledge structures that can be used for dealing with the environment in suitable ways. These knowledge structures constitute a core that supports several cognitive processes, such as perception, categorization, planning, etc. The Artificial Intelligence, as a research field, aims at developing approaches for mimicking these cognitive capabilities in machines. Due to this, it is important to investigate approaches that allow representing the knowledge in flexible ways. In order to overcome some limitations of the classical theory of knowledge representation, which is adopted by several approaches proposed in the Artificial Intelligence field, this work proposes a cognitively-inspired framework for knowledge representation and reasoning which integrates aspects from three different cognitive theories about concept representation in the human cognition: classical theory, prototype theory and exemplar theory. The resulting framework can support compositionality, typicality, representation of atypical instances of concepts, and representation of the variability of the individuals classified by each concept. Consequently, the proposed framework also supports logical reasoning and similarity-based reasoning. The main contributions of this work are the formalization of a cognitively-inspired framework for knowledge representation and reasoning, two approaches for selecting representative exemplars of each concept and an approach of reasoning for classification that integrates logical reasoning and similarity-based reasoning and that is supported by definitions, prototypes and exemplars of concepts. This thesis also presents a system for automatic interpretation of depositional processes application that adopts the proposed framework. The experiments, which were performed on a classification task, suggest that the proposed framework provides classifications that are more informative than the ones provided by a classical approach.

The rapid growth of sensor data can potentially enable a better awareness of the environment for humans. In this regard, interpretation of data needs to be human-understandable. For this, data interpretation may include semantic annotations that hold the meaning of numeric data. This thesis is about bridging the gap between quantitative data and qualitative knowledge to enrich the interpretation of data. There are a number of challenges which make the automation of the interpretation process non-trivial. Challenges include the complexity of sensor data, the amount of available structured knowledge and the inherent uncertainty in data. Under the premise that high level knowledge is contained in ontologies, this thesis investigates the use of current techniques in ontological knowledge representation and reasoning to confront these challenges. Our research is divided into three phases, where the focus of the first phase is on the interpretation of data for domains which are semantically poor in terms of available structured knowledge. During the second phase, we studied publicly available ontological knowledge for the task of annotating multivariate data. Our contribution in this phase is about applying a diagnostic reasoning algorithm to available ontologies. Our studies during the last phase have been focused on the design and development of a domain-independent ontological representation model equipped with a non-monotonic reasoning approach with the purpose of annotating time-series data. Our last contribution is related to coupling the OWL-DL ontology with a non-monotonic reasoner. The experimental platforms used for validation consist of a network of sensors which include gas sensors whose generated data is complex. A secondary data set includes time series medical signals representing physiological data, as well as a number of publicly available ontologies such as NCBO Bioportal repository.

These are the proceedings of the International Workshop on Reactive Concepts in Knowledge Representation (ReactKnow 2014), which took place on August 19th, 2014 in Prague, co-located with the 21st European Conference on Artificial Intelligence (ECAI 2014).