Ing. Martin Ledvinka, Ph.D.

Domain-independent information systems like ontology editors provide only limited usability for non-experts when domain-specific linked data need to be created. On the contrary, domain-specific applications require adequate architecture for data authoring and validation, typically using the object-oriented paradigm. So far, several software libraries mapping the RDF model (representing linked data) to the object model have been introduced in the literature. In this paper, we develop a novel framework for comparison of object-triple mapping solutions in terms of features and performance. For feature comparison, we designed a set of qualitative criteria reflecting object-oriented application developer's needs. For the performance comparison, we introduce a benchmark based on a real-world information system that we implemented using one of the compared object-triple mapping (OTM) solutions -- JOPA. We present a detailed evaluation of a selected set of OTM libraries, showing how they differ in terms of features. We further narrow down the set of selected OTM libraries to contain only RDF4J-compatible ones and use the benchmark to measure their time and memory efficiency.

Ontology learning from a text written in natural language is a well-studied domain. However, the applicability of techniques for ontology learning from natural language texts is strongly dependent on the characteristics of the text corpus and the language used. In this paper, we present our work so far in entity linking and enhancing the ontology with extracted relations between concepts. We discuss the benefits of adequately designed lexico-semantic patterns in ontology learning. We propose a preliminary set of lexico-semantic patterns designed for the Czech language to learn new relations between concepts in the related domain ontology in a semi-supervised approach. We utilize data from the urban planning and development domain to evaluate the introduced technique. As a partial prototypical implementation of the stack, we present Annotace, a text annotation service that provides links between the ontology model and the textual documents in Czech.

Many organizations already benefit from using semantic vocabularies which help them systematize, search and reuse their data. However, to efficiently manage such vocabularies, appropriate and adequate tools are needed. In this paper, we present TermIt, an integrated system for managing a set of interconnected vocabularies, identification of individual concepts in source documents, and using such terms for semantic data asset annotation and subsequent search. We relate TermIt to other relevant tools and present usage scenarios we have identified so far.

Ontologies can represent a significant asset of domain-specific information systems, written predominantly using the object-oriented paradigm. However, to be able to work with ontological data in this paradigm, a mapping must ensure transformation between the ontology and the object world. While many software libraries provide such a mapping, they lack standardization or formal guarantees of its semantics. In this paper, we provide a formalism for mapping ontologies between description logics and F-logic, a formal language for representing structural aspects of object-oriented programming languages. This formalism allows to precisely specify the semantics of the object-ontological mapping and thus ensure a predictable shape and behavior of the object model.

Using ontologies for enterprise data integration brings an opportunity to use them for efficient data query evaluation. The rationale is that ontologies represent common-sense structures that are reflected also in queries posed by data users and thus can be used for query optimization. This paper presents how proper foundational-ontology-based knowledge can be used to design a generic index, which can help in answering a wide range of queries compliant with the foundational ontology. We discuss several indexing techniques and evaluate our proposal for different UFO-based queries extracted from different query sets.

Creating applications on top of linked data and ontologies brings many difficulties. The applications are either generic (and thus not appealing to end-users), or bound to ontology structure, change of which breaks the application. We present JOPA, a tool that formalizes the contract between the application and the ontology, combining advantages of both worlds. JOPA is a persistence framework for Java applications, providing formalized object-ontological mapping, transactions, access to multiple repository contexts, and producing linked data. The system is demonstrated on a real use-case of a reporting tool that we develop for the aviation industry.

Accessing OWL ontologies programmatically by complex IT systems brings many problems stemming from ontology evolution, their open-world nature and expressiveness. This paper presents Java OWL Persistence API (JOPA), a persistence layer that allows using the object-oriented paradigm for accessing semantic web ontologies. Comparing to other approaches, it supports validation of the ontological assumptions on the object level, advanced caching, transactional approach, unification and optimization of repository access through the OntoDriver component, as well as accessing multiple repository contexts at the same time. Additionally, we present a complexity analysis of OntoDriver operations that allows optimizing objectoriented access performance for underlying storage mechanisms. Last but not least, we compare our object-oriented solution to low level Sesame API in terms of efficiency.

Accessing OWL ontologies from IT systems can bring many problems unfamiliar to developers used to the more common relational storage approach. These problems stem from the dynamic nature of ontologies, their open-world character and expressiveness. In this paper, we present the Java OWL Persistence API (JOPA), a persistence layer allowing object-oriented access to semantic web ontologies. It supports features like caching, transactional processing and a semantically clear contract between the ontology and the object model. In addition, we present the OntoDriver, a software layer decoupling storage access from the object-ontological mapping. We provide an in-depth theoretical complexity analysis of our approach in connection with an analysis and practical evaluation of the performance of ontological storage with regards to application access scenario.

Although many applications built on top of market-ready ontology storages are generic and lack dependence on the particular application domain, most users prefer applications tailored to their particular task. Such applications are typically built using object-oriented paradigm that accesses data differently than generic applications. In this paper, we define a benchmark consisting of an ontology and ontological queries tailored for testing suitability of ontological storages for object-oriented access. We present results of experiments on several state-of-the-art ontological storages and discuss their suitability for the purpose of object-oriented application access.

Accessing OWL ontologies programmatically by complex IT systems brings many problems stemming from ontology evolution, their open-world nature and expressiveness. This paper introduces Java OWL Persistence API (JOPA) that tackles these problems by providing Object-Ontological Mapping (OOM) driven by integrity constraints designed on top of OWL ontologies, together with transactional processing and various backend implementations. Based on the lessons learnt during the design and implementation of the prototype, we present a proposal for a two layered architecture of object-oriented ontological information system design and analyse complexity of various operations for object-oriented ontological applications.

The MONDIS project is aimed at knowledge management in the field of monument conservation, monument damages and related interventions. The IT platform developed within the project is to be used by conservators, monument owners and other stakeholders to share their terminology, contextual knowledge, as well as experience about existing damages and their interventions.