Ing. Petr Křemen, Ph.D.

Recently, there has been growing interest in the use of ontology as a fundamental tool for representing domain-specific conceptual models to improve the semantics, accuracy, and relevance of domain users’ query results. Although the amount of data has grown steadily over the past decade, much data shares similar characteristics that can be captured by a foundational ontology. In this paper, we show how queries based on a foundational ontology can be evaluated and their performance measured. We also present a Foundational Patterns benchmark to help select the most efficient triple memory and its layout. We evaluate the foundational benchmark with both generated and real datasets for state-of-the-art triple stores.

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.

Aviation is a high risk industry where safety management is mandatory for organizations. On a global scale, safety management is hierarchical, i.e. national authorities manage safety of the sector, controlling organizations which in turn manage safety of their operations, as well as the safety of their organizational units. It is apparent that safety information management is a key factor to the success of safety management. Improving safety information management in aviation can be achieved by the systematic documentation of safety information and its seamless sharing vertically as well as horizontally through the management hierarchy. Despite attempts to tackle this problem, the industry still suffers from information management issues, e.g. usage of terminologically incompatible safety management frameworks. As a result diverse information repositories, e.g. investigation reports, safety recommendations, or audit reports managed by these frameworks, cannot be efficiently explored and compared. We apply Linked Data principles, e.g. searchability and explorability based on dataset descriptors, to help solving the problem. In this work we propose an ontology-based safety information management for the aviation industry. To achieve this we develop the Aviation Safety Ontology that is modular and covers the contexts of the organization types aerodrome, airline, air traffic management, maintenance and the state civil aviation authority. We test the appropriateness of the proposed ontology by representing safety data in terms of the ontology.

The primer focus of the paper is to introduce a new concept of aviation industry approach to the risk management issues and following control activities. Due to its size, complexity and dynamic changes, which take place almost continuously, a platform was developed with one important goal, to help in creation of the industry-wide approach to the field of safety management. The approach is recognised as potentially beneficial for many entities involved. Such integration would enable a definition of the three aspects, which were introduced and further analysed - intra organizational, inter organizational and holistic aviation data approach. Each covers different layers or structure parts whose relations and definitions are both relevant and logical. All aspects represent different points of view, starting from the organisation level and leading to the field of the mutual activities involving various industry stakeholders and relevant authorities. Expected solution should ensure standardised approach to the reporting and following investigation. Ontological modelling would support an integration and development of the mutual risk assessment and control mechanisms. Building standardised aviation ontology, developed through one mutual system emerges as potentially powerful for application of various analytics instruments, e.g. business intelligence.

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.

Querying geospatial data from multiple heterogeneous sources backed by different management technologies poses an interesting problem in the data integration and in the subsequent result interpretation. This paper proposes broker techniques for answering a user’s complex spatial query: finding relevant data sources (from a catalogue of data sources) capable of answering the query, eventually splitting the query and finding relevant data sources for the query parts, when no single source suffices. For the purpose, we describe each source with a set of prototypical queries that are algorithmically arranged into a lattice, which makes searching efficient. The proposed algorithms leverage GeoSPARQL query containment enhanced with OWL 2 QL semantics. A prototype is implemented in a system called OnGIS.

Last years, there has been growing interest in developing high quality models to ensure interoperability of applications as well as proper understanding among a community. To improve productivity of model designers as well as to improve quality of resulting models, proper theoretical foundations and tool support is necessary. This paper discusses some types of conceptual modeling anti-patterns that lead to error- prone modeling decisions, and describes a reproducible solution to a general anti-pattern detection problem during conceptual design. Novel contribution of this paper is the definition of new ontological anti-patterns, we observed during our work in design and model ontologies in the domain of Aviation Safety. The approach is illustrated on models designed by means of OntoUML, an ontology-founded UML profile based on the Unified Foundational Ontology (UFO).

In recent decades, in the aviation safety community, there is an increasing amount of requirements especially relating to technology and performance. In this paper we discuss how to model aviation safety domain using Unified Foundational Ontology (UFO), which is a top-level ontology aimed at specifications of domain ontologies, which are formal descriptions of concepts and the relationships between these concepts that describe a specific application area. UFO can be used for evaluating business modeling methods and providing real-world semantics for their modeling constructs.

Datasets are created in the linked data ecosystem with varying data and metadata quality. Shallow (or poor-quality) metadata prevent dataset provision services (CKAN catalogs, etc.) from supporting their users in efficient search and exploration of datasets they need. In this paper we approach the problem by formalizing the dataset exploration scenario, introducing the notion of ontology-based dataset descriptors. We present a particular use-case of creating temporal dataset descriptors and their application in the field of Czech Open Data cloud. We also discuss practical problems that we came across during descriptor extraction. Experimental datasets and computed descriptors are published in a public SPARQL endpoint.

Aviation safety reports start to play a crucial role in understanding incidents and accidents in the aviation safety field. Automated text processing is necessary for simplification of the safety reporting process.This task can be acheived in diffe-rent ways, such as statistical, non-statistical or a combination of these techniques. In this paper, we are mainly focusing on non-statistical ones, by introducing our text processing scenario. We start with indexing of various aviation safety vocabularies that we are using as a backbone for this task. Next, the golden stan-dard corpus is prepared, including the testing process of several Linked Data Knowledge Extraction tools, with respect to a domain-specific vocabulary. Then, choosing the most accurate entity annotation tools and making them work toget-her, as well as with other features that we added, taking into consideration some very specific terms and abbreviations used in aviation field. The ultimate goal is to build a tool that will integrate several techniques inside in order to provide high precision reports’ annotations in aviation safety domain.

In recent years, human-ontology interaction becomes an increasingly important subject for computational and information systems developers. Human information consumers and web agents need to use and query ontologies using their web sites and web applications, thus the need for developing tool supporting ontological engineering and querying tools arises. In this paper we discuss the potential of web based human-readable ontology queries. In large taxonomies such Aviation Safety (AS) domain, it is important to allow easier navigation within the ontology. Thus we introduce an extension to the OntoQuery tool for more practical visualization of query results and easy OWL vocabulary dissemination to the community.

As the size of semantic data available as Linked Open Data (LOD) increases, the demand for methods for automated exploration of data sets grows as well. A data consumer needs to search for data sets meeting his interest and look into them using suitable visualization techniques to check whether the data sets are useful or not. In the recent years, particular advances have been made in the field, e.g., automated ontology matching techniques or LOD visualization platforms. However, an integrated approach to LOD exploration is still missing. On the scale of the whole web, the current approaches allow a user to discover data sets using keywords or manually through large data catalogs. Existing visualization techniques presume that a data set is of an expected type and structure. The aim of this position paper is to show the need for time and space efficient techniques for discovery of previously unknown LOD data sets on the base of a consumer’s interest and their automated visualization which we address in our ongoing work.

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.

Evolving ontologies by domain experts is difficult and typically cannot be performed without the assistance of an ontology engineer. This process takes long time and often recurrent modeling errors have to be resolved. This paper proposes a technique for creating controlled ontology evolution scenarios that ensure consistency of the possible ontology evolution and give guarrantees to the domain expert that his/her updates do not cause inconsistency. We introduce ontology templates that formalize the notion of controlled evolution and define ontology template consistency checking service together with a consistency checking algorithm. We prove correctness and demonstate the practical use of the techniques in two scenarios.

UML and OWL have different goals and approaches. However they have some similarities, especially for representing structure (class diagrams). In this paper we tried to develop OWLtoUML tool to visualize OWL in UML. This Tool can be useful to transform owl ontology to UML class that can be helpful to present the data structure and organization.

Deriving from the complex nature of cultural heritage conservation and its heterogeneous tasks it is the need for enhancing a systematic but flexible organization of expert knowledge in the field. Such organization should address comprehensively the interrelations and complementariness among the different factors that come into play in the understanding of diagnostic and intervention problems. The purpose of the system introduced in this paper, called MONDIS, is to endorse this kind of organization. The system allows replicating in a computer readable form the basic dependence among factors influencing the description, diagnosis and intervention of damages to immovable objects. More specifically MONDIS allows to input and search entries concerning object description, structural evolution, location characteristics and risk, component, material properties, surveys and measurements, damage typology, damage triggering events and possible interventions. The system supports searching features typical of standard databases, as it allows for the digitalization of a wide range of information including professional reports, books, articles and scientific papers. It also allows for computer aided retrieval of information tailored to user's requirements. The approach consists in formalising an ontological representation of the field of heritage conservation compatible with the professional approach commonly employed to processing of data. The paper also presents a series of applications which run on such ontology, including a web-based interface for data inputting, a terminology editor for defining and searching a pictographic glossary of all terms stored in the system, a knowledge matrix for browsing the cause-effect relationships in damage formation together with insights on possible interventions and a mobile application for supporting in-situ condition assessment of buildings.

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.

In this paper we study the distinct count aggregation function used in queries into expressive ontologies. The main differences in this settings opposed to aggregation in relational database systems are the Open World Assumption and incomplete knowledge. We propose different interpretations useful in different practical use-cases of the distinct count function, i.e. basic count, semantic count, epistemic count and semantic tuple count some of which use knowledge derived from the ontology in order to obtain results in accordance with the Open World Assumption. We use interval semantics to model the uncertainty of the distinct count function's result induced by incomplete knowledge in the ontology. We show that interval semantics are particularly useful in aggregate queries with filtering clause and when we need the retrieval of the boundaries of the uncertainty of the distinct count function's result. We study and present relationships among the different interpretations and decidability of the semantic tuple count. We also propose a theoretical approximation of the semantic tuple count. Our results are applicable to a wide range of description logic formalisms allowing to express equality/inequality between individuals, concepts and relations, e.g.Web Ontology Language.

The activities involved in value chain based production have become more complex with an intreasing amount of interdependencies, new technologies and materials that introduce new risks in an ever changing environment. At the same time our society has become more risk averse. Until the start of the IRIS project in 2008 very little had been done to identify industrial risk factors in a systematic manner and risk assessment and management was largely characterized by its methodical diversity and fragmented approach. A shift to a new safety paradigm was therefore requied in order to combine a more competitive, innovative and better risk informed industry with a safer cosiety in mind.

Deriving from the complex nature of cultural heritage conservation it is the need for enhancing a systematic but flexible organization of expert knowledge in the field. Such organization should address comprehensively the interrelations and complementariness among the different factors that come into play in the understanding of diagnostic and intervention problems. The purpose of MONDIS is to endorse this kind of organization. The approach consists in applying an ontological representation to the field of heritage conservation in order to establish an appropriate processing of data. The system allows replicating in a computer readable form the basic dependence among factors influencing the description, diagnosis and intervention of damages to immovable objects. More specifically MONDIS allows to input and search entries concerning object description, structural evolution, location characteristics and risk, component, material properties, surveys and measurements, damage typology, damage triggering events and possible interventions. The system supports searching features typical of standard databases, as it allows for the digitalization of a wide range of information including professional reports, books, articles and scientific papers. It also allows for computer aided retrieval of information tailored to user's requirements. The foreseen outputs will include a web user interface and a mobile application for visual inspection purposes.

Kapitola Ontologie a deskripční logiky představuje logický aparát použivaný v současných technologiích sémantického webu. Čtenář se seznámí se základním jazykem ALC a tablovým algoritmem používaným pro ověřování konzistence ontologie v tomto jazyku. V další části kapitoly jsou představeny pokročilé odvozovací procedury, a to vysvětlování modelovacích chyb a metody pokládání komplexních dotazů.

Web Ontology Language ontologies become more and more popular in complex domain modeling for their high expressiveness, flexibility and well defined semantics. Although query languages adequate in expressiveness to OWL reasoning capabilities were introduced before, their implementations are rather limited. In this paper, we study SPARQL-DL^NOT, an extension of one of these query languages, SPARQL-DL, and present novel evaluation and optimization techniques for efficient SPARQL-DL^NOT execution. As queries become complex easily, we also present a novel graph-based visualization that simplifies query construction and maintenance. Presented techniques and algorithms were implemented in the Pellet reasoner and in our novel Protégé plug-in OWL2Query.

Capturing knowledge about damages and failures of culture heritage objects is a complex task because of term ambiguity, knowledge incompleteness and variety. To tackle this complexity, the paper introduces a semantic web ontology that aims at modeling monument damage knowledge with significant contextual information, including monument identification, damage identification, risk assessment, damage diagnosis and remedial measures. The developed ontology is being tested in the MONDIS project as a background knowledge for custom software tools for management of damage failure knowledge. During the course of the project it becomes a common model for linked-data-compliant knowledge-based system serving different stakeholders to model/retrieve/compare different cases of damage and successful interventions.

Conventional information systems are built on top of a relational database that requires its data model to be stable. This lack of adaptability is very restrictive for systems manipulating evolving or heterogenous knowledge. The authors of this paper faced this issue when developing an information system of structural failures. As knowledge on structural failures is very heterogeneous and dynamic, it is practically impossible to define a stable database schema ahead. The widely accepted alternative to relational databases are semantic web ontologies. However, software developers miss adequate frameworks and design methodologies in this area. This paper tries to fill the gap by proposing a methodology for designing ontology-backed software applications that makes the ontology possible to evolve while being exploited by one or more applications at the same time. A reference Java implementation and the proof-of-concept application is introduced at the end of the paper.

Mindmaps have been a popular means for intuitive and user-friendly conceptual modeling for several years. This paper proposes a mindmapping tool driven by semantic web ontologies that allows to assign meanings to the mindmap nodes and, even more importantly, guide the user through mindmap creation by proposing ontological concepts and roles relevant to each mindmap node. The tool is backed by OWL ontologies that allow to capture domain knowledge semantics precisely and rigorously. The tool is being developed as a part of the MONDIS project aimed at efficient knowledge management of immovable cultural heritage failures records. As such, the tool is currently used by civil engineering experts for collaborative authoring of structural damage records.

High expressiveness of semantic web query languages, like SPARQL or SPARQL-DL, brings practical challenges in query author- ing and visualization for both experts and non-experts. In this paper we present the successful OWL2Query plug-in for Protégé, a simple and feature rich graphical visualization and editing tool for complex OWL 2 queries expressed in SPARQL-DL query language. Our tool is designed and implemented with the goal of inspiring and supporting users in query authoring, visualization and evaluation. The interactive query editing and the intuitive visualization are not only helping users in quick author- ing of valid queries, but also are crucial points in the efficient adoption of semantic technologies by non-experts.

The problem addressed in this paper is the challenge of automated construction of knowledge discovery workflows, given the types of inputs and the required outputs of the knowl- edge discovery process. Our methodology consists of two main ingredients. The first one is defining a formal conceptualization of knowledge types and data mining algorithms by means of knowl- edge discovery ontology. The second one is workflow composition formalized as a planning task using the ontology of domain and task descriptions. Two versions of a forward chaining planning algorithm were developed. The baseline version demonstrates suitability of the knowledge discovery ontology for planning and uses Planning Domain Definition Language (PDDL) descriptions of algorithms; to this end, a procedure for converting data mining algorithm descriptions into PDDL was developed. The second directly queries the ontology using a reasoner.

Conjunctive query answering is becoming a very important task on the Semantic Web as the adoption of SPARQL query language increases. There is considerable work done in the area of optimizing conjunctive query answering for RDF and OWL2-DL ontologies, in the latter case namely for queries without undistinguished variables. However, there has not been much emphasis on how to handle queries with both distinguished and undistinguished variables efficiently. In this paper, we present a novel algorithm for answering conjunctive queries over OWL2-DL ontologies with undistinguished variables efficiently. These optimizations are implemented in the Pellet query engine and our experimental evaluation shows that the new optimization improves the query performance significantly.

This document describes a knowledge base of structural failures StruFail. The system is backed by OWL ontologies and allows to author records of failures of immovable objects and pose expressive queries. As a part of the system, OWL ontologies describing failures, objects, materials, and other domain terms have been developed. The system is described from the domain expert perspective, introducing the main features and design decisions as well as the way how to interact with the system.

A particular passive decision support system (VCDECIS) uses Topic Maps as a semantic technology for representing its knowledge. A decision was taken to transform the knowledge representation to Web Ontology Language, mainly to get the additional opportunity to use more reasoning features. In this contribution we show how we could overcome the shortcomings of the Topic-Map representation by applying our developed transformation concept for the particular system VCEDECIS.

In this contribution we show, how we could overcome the shortcomings of the Topic-Map representation by applying our developed transformation concept for the particular system VCEDECIS. At a passive decision support system using Topic Maps as a semantic technology for representing its knowledge, it was decided to transform the knowledge representation to Web Ontology Language. We introduce a transformation concept that starts with a sound analysis of the source system. For typical patterns in the topic map the best corresponding patterns in OWL-DL were defined. A combination of general considerations and real examples act as a proof of concept.

As the semantic web is becoming more mature, the need for complex, large, formally described ontologies grows. Developing such ontologies is rather nontrivial and complex task requiring collaboration among knowledge engineers and domain experts. To support ontology sharing we introduce OWLDiff, an OWL 2 ontology comparison/merge tool, its algorithms, implementation details and usage examples. OWLDiff can be used for syntactic, explanation-based and semantic comparison and merging of OWL 2 ontologies and integrates also a Subversion client for ontology versioning. The system is open-source and is available as a standalone application, Prot'eg'e plugin and NeOn toolkit plugin.

If home technologies are to be successful in supporting independence of older people, they should be designed with the needs of all users involved in the service delivery in mind. Thus, involving potential users from the very beginning and in all phases of the development process is crucial when designing Ambient Assisted Living systems and services. This paper describes experiences from two European research projects in making user-centred design a reality.

The road towards implementing technology and services developed in research projects is often a bumpy one. Developing in a controlled lab environment requires another type of expertise than implementing technology in real-life situations. However when the two types of expertise are combined during the research phase they can benefit from each other and achieve good results. For that other types of skills, such as communicative skills and expertise in working in a multidisciplinary team are also required. This paper gives insight into the approach taken in the Netcarity project to being able to deliver services in real homes.

This paper presents a novel semantic framework for authoring privacy policies of servers and privacy preferences of clients as well as their matching. The proposed solution allows for interactive detection and explanation of semantic inconsistencies during policy/preferences design. Privacy policies are expressed as OWL-DL ontologies and privacy preferences as OWL-DL integrity constraints checked via SPARQL-DL queries. The well-defined semantics of our solution together with flexibility of semantic technologies fixes known serious problems of P3P and APPEL matching scenario, including ambiguity and semantical incorrectness.

Explanations of modeling errors in ontologies are of crucial importance both when creating and maintaining the ontology. This work presents two novel incremental methods for error explanations in semantic web ontologies and shows their advantages w.r.t. the state of the art black-box techniques. Both promising techniques together with our implementation of a tableau reasoner for an important OWL-DL subset SHIN are used in our semantic annotation tool prototype to explain modeling errors.

One of the interesting problems related to semantic technologies is ontology versioning. In this case simple text-based comparison of versions is logically insufficient. Differences between versions have to be considered according to its semantics, but it is also useful to have a tool that manages standard versioning repository and usual files in a traditional way (like Subversion). In this paper we introduce a system for ontology versioning which is able to work with ontology files (modelled using OWL) from the repositories and provides both standard and semantic diff/merge operations. The whole system (implemented under open-source license) called OWL Version (or OVN) will be based on standard Subversion technology and existing OWLDiff package, which will be enhanced in order to achieve functional and technological requirements. Current status of the project is work-in-progress. The finished project will also produce versioning support available as Protégé plugin.

Compared to traditional ways of annotating multimedia resources (textual documents, photographs, audio/video clips etc.) by keywords in form of text fragments, semantic annotations are based on tagging such multimedia resources with meaning of objects (like cultural/historical artifacts) the resource is dealing with. The search for multimedia resources stored in a repository enriched with semantic annotations makes use of an appropriate reasoning algorithm. Knowledge management and Semantic Web communities have developed a number of relevant formalisms and methods. This chapter is motivated by practical experience with authoring of semantic annotations of cultural heritage related resources/objects.

P3P is a W3C recommendation for expressing service privacy practices in a structured computer-readable form. However, the lack of formal semantics makes it inconvenient for fully computerized matching against privacy preferences (e.g. the APPEL language). SemPref is an interesting approach that defines several possible semantics for P3P using relational database technology. However, it still doesn't have capabilities to detect errors that frequently occur when designing both privacy preferences and service privacy practices. This paper analyses the problem of detecting semantic errors in P3P policies in depth and proposes a solution based on semantic technologies.P3P is a W3C recommendation for expressing service privacy practices in a structured computer-readable form. However, the lack of formal semantics makes it inconvenient for fully computerized matching against privacy preferences (e.g. the APPEL language).

This paper addresses the problem of semi automatic design of workflows for complex knowledge discovery tasks. Assembly of optimized knowledge discovery workflows requires awareness of and extensive knowledge about the principles and mutual relations between diverse data processing and mining algorithms. We aim at alleviating this burden by automatically proposing workflows for the given type of inputs and required outputs of the discovery process. The methodology adopted in this study is to define a formal conceptualization of knowledge types and data mining algorithms and design a planning algorithm, which extracts constraints from this conceptualization for the given user's input output requirements. We demonstrate our approach in two use cases, one from scientific discovery in genomics and another from advanced engineering

Recently, SPARQL-DL was introduced as a rich query language for OWL-DL ontologies. It provides an OWL-DL-like semantics for SPARQL basic graph patterns which involves as special cases both conjunctive ABox queries and mixed TBox/RBox/ABox queries over Description Logic (DL) ontologies. This paper describes the implementation of a SPARQL-DL query engine and discusses several optimizations. We investigate the new challenges brought by the additional expressivity of SPARQL-DL by extending a well-known benchmark.

This paper addresses the problem of semi automatic design of workflows for complex knowledge discovery tasks.

Incremental approach to error explanations in ontologies are of crucial importance both when creating and maintaining the ontology.

"Efficient searching in large document repositories needs semantic annotations that are both expressive enough to allow interesting queries and weak enough to enable efficient reasoning. Lack of usable automatic procedures for creating such annotations drives the research in algorithms and methods for semi-automatic knowledge base building. In this work, we address the problem of error explanations and debugging. Two novel incremental methods for error explanations are introduced and compared to the state of the art black-box techniques. Both promising incremental techniques together with our implementation of a tableau reasoner for SHIN logic are used to explain modeling errors in annotations authored with our annotation tool prototype."

Relational Data Mining Applied to Virtual Engineering of Product Designs

Contemporary product design based on 3D CAD tools aims at improved effciency using integrated engineering environments with access to databases of existing designs, associated documents and enterprize resource planning. The project SEVENPRO addresses this issue by developing a framework for integration of the different information sources through a layer of semantic annotations and tools for enhancing management and reuse of engineering knowledge. The discovery of implicit knowledge by means of relational data mining is expected to be an important facility for reusing of engineering knowledge. Therefore we are developing a system for relational data mining, which could be seamlessly integrated into the semantic engineering environment.

This paper applies the methodology of subgroup discovery which first seeks frequent structural patterns in CAD designs and then discovers interesting (large and class-biased) design subgroups defined in terms of the frequent patterns its members contain. Besides the two mentioned kinds of data mining results, further categories of patterns will be required to mine in the present application domain. The large amount of generated results of each kind calls for establishing an efficient pattern management and search functionality. For this purpose we elaborate an ontology of relational data mining results.

The paper suggests a technique for searching temporal patterns, which is tested on the hepatitis data set. Relevant preprocessed time series is divided into several periods with respect to the given significant instants (for example biopsy date or start of interferon application), development trends are computed. Then dependencies in form of complex association rules between these groups are searched for. We have focused on the analysis of differences between examination results before and after interferon therapy for hepatitis C patients. To reach this goal two software systems SumatraTT and LISp-Miner are combined -- sophisticated data transformations and enhancements are designed and ensured through SumatraTT while LISp-Miner takes care for the search of interesting significant differences in the preprocessed datasets.

Contemporary product design based on 3D CAD tools aims at improved efficiency using integrated engineering environments with access to databases of existing designs, associated documents and enterprise resource planning (ERP). One of the goals of the SEVENPRO project is to achieve design process improvements through the utilization of relational data mining (RDM), utilizing past designs and commonly agreed design ontologies. In this paper we discuss the applicability of state-of-the-art ILP systems to the RDM tasks in this application domain as well as the implied challenges to ILP research and system implementations.

This poster presents our proposal to graphical interface to semantic annotations, implemented in the DNA tool. The system uses conceptual graphs as a suitable compromise between understandability and formality.

The paper suggests a methodology for search of temporal patterns, which is tested on the problem of difference between hepatitis B and C. To reach this goal two software systems LISp-Miner and SumatraTT are combined - sophisticated data transformations and enhancements are designed and ensured through SumatraTT while LISp-Miner takes care for the search of significant interesting differences in the resulting datasets. The main obtained results are reviewed in the section 3 - there are identified some examinations the values of which significantly differ for both types of attributes. This proves that the suggested general methodology has promising potential when applied to the considered type of data. A plan for additional data-mining questions to be studied later is presented in the Conclusions.