Ontology Service Center: A Datahub for Ontology Application

With the growth of data-oriented research in humanities, a large number of research datasets have been created and published through web services. However, how to discover, integrate and reuse these distributed heterogeneous research datasets is a challenging task. Ontology is the soul between series digital humanities resources, which provides a good way for people to discover and understand these datasets. With the release of more and more linked open data and knowledge bases, a large number of ontologies have been produced at the same time. These ontologies have different publishing formats, consumption patterns, and interactions ways, which are not conductive to the user’s understanding of the datasets and the reuse of the ontologies. The Ontology Service Center platform consists of Ontology Query Center and Ontology Validation Center, mainly using linked data and ontology-based technologies. The Ontology Query Center realizes the functions of ontology publishing, querying, data interaction and online browsing, while the Ontology Validation Center can verify the status of using certain ontologies in the linked datasets. The empirical part of the paper uses the Confucius portrait as an example of how OSC can be used in the semantic annotation of images. In a word, the purpose of this paper is to construct the applied ecology of ontology to promote the development of knowledge graphs and the spread of ontology.

High-Level Smart Decision Making of a Robot Based on Ontology in a Search and Rescue Scenario

The search and rescue (SAR) scenario is complex and uncertain where a robot needs to understand the scenario to make smart decisions. Aiming at the knowledge representation (KR) in the field of SAR, this paper builds an ontology model that enables a robot to understand how to make smart decisions. The ontology is divided into three parts, namely entity ontology, environment ontology, and task ontology. Web Ontology Language (OWL) is adopted to represent these three types of ontology. Through ontology and Semantic Web Rule Language (SWRL) rules, the robot infers the tasks to be performed according to the environment state and at the same time obtains the semantic information of the victims. Then, the paper proposes an ontology-based algorithm for task planning to get a sequence of atomic actions so as to complete the high-level inferred task. In addition, an indoor experiment was designed and built for the SAR scenario using a real robot platform—TurtleBot3. The correctness and usability of the ontology and the proposed methods are verified by experiments.

Towards Task Understanding in Visual Settings

We consider the problem of understanding real world tasks depicted in visual images. While most existing image captioning methods excel in producing natural language descriptions of visual scenes involving human tasks, there is often the need for an understanding of the exact task being undertaken rather than a literal description of the scene. We leverage insights from real world task understanding systems, and propose a framework composed of convolutional neural networks, and an external hierarchical task ontology to produce task descriptions from input images. Detailed experiments highlight the efficacy of the extracted descriptions, which could potentially find their way in many applications, including image alt text generation.

Roles of Ontologies in Cartography

<p><strong>Abstract.</strong> Ontologies (in computer science and information science) represent the essential tool for a formalised description of concepts, data, information, knowledge and other entities as well as relations among them. Their history is relatively old. The idea of ontologies in informatics started in the mid-1970s, but ontology as the philosophical discipline connected to existence and nature of reality came from the Ancient Greek. The ontologies as a part of knowledge-based systems were discussed in the 1980s. In 1993 Thomas R. Gruber defined ontology in information science as "a specification of a conceptualisation". After that, the first languages and formats coding ontologies have been developed, and massive construction process of ontologies began. For example, the Basel Register of Thesauri, Ontologies and Classifications presents about 700 ontologies and more the 1000 other tools with a similar character. The theory of ontologies and development as ontologies are entirely on a high level. However, their implementation (especially in several domains) is in its infancy.</p><p> For example, in the geographical domain, there are many ontologies (called geo-ontologies) such as FAO (Food and Agriculture Organization of the United Nations) Geopolitical Ontology, ontologies of USGS (United States Geological Survey) or ontologies of Ordnance Survey. However, their implementation is usually limited by home organisations, which provide for the management, development and updating of ontologies. In many cases, they are not an integral part of Linked Open Data (LOD). This fact can be considered as the critical shortcoming because only in connection with Linked Open Data and free data sharing and combining the main benefits of ontologies (emphasis on a semantic description, derivation of new knowledge or complete independence) can be fully appreciated.</p><p> This document has to describe opportunities for the implementation of ontologies in cartography. The purpose of the implementation of an ontology depends on various types of ontologies. There are defined four essential types of ontologies - upper ontologies, domain ontologies, task ontologies and application ontologies.</p><p> Upper and domain ontologies contain general terms (in the case of upper ontologies) and domain-specific terms (in case of domain ontologies). Annotation properties (labels, definitions or comments) usually describe these terms, interconnected by data properties and/or object properties and restricted by logical axioms. Such ontologies are usually provided as vocabularies or thesauri. They can be used in two ways. Domain ontologies can describe cartography as a science or human activity. In previous years several paper and articles were discussing the term "cartography" and its position in Linked Open Data space, including various ontologies, ontological description of cartographic knowledge or ontological comparison of various definitions of the term "map". These activities can aim for the development of a cartographic knowledge base or building of semantic tools such as multilingual thesauri or vocabularies.</p><p> The second way consists in the exploitation of domain ontologies containing semantic information about data visualising by a map. In this case, such domain ontology can be used as a tool for development of a legend of a map, especially in a case where a map is focused on particular issues. If such ontology is published as Linked Open Data, it is possible to generate such legend automatically as well as to reflect any changes. Such solution enables an efficient interconnection of cartographers and domain experts. Domain ontologies can be used for a definition of logical rules restricting and describing data, information and knowledge. These rules and knowledge extracted in the reasoning process can be applied during the map development. They can provide information on possible combinations of data or a hierarchy of objects visualising by a map and described by a map legend.</p><p> The task ontologies are not focused on a complicated system of classes (representing types of object) as domain ontologies. They are usually based on instances (individuals) representing concrete data objects. Therefore they can be used as data resources. However, the overwhelming majority of geo-ontologies does not contain any geometry (coordinates) to enable a visualisation in a map. This apparent disadvantage shows the importance of LOD. If a task ontology is published as 5-star LOD (RDF /Resource Description Framework/ data with interconnection to external data resources published on the Web under an open license), and identity relation (links to equivalent object published in other data sets) are filled, it is possible to find in LOD space geometries as well as other additional information and attributes for visualization.</p><p> The remaining type of ontologies is called application ontology. It is a combination of both previous kinds &amp;ndash; domain ontology and task ontology. Application ontologies usually provide vocabularies as well as data stored in an ontological structure. Such a combination allows controlling data correctness and integrity by a set of logical rules. This functionality is emphasised by the rich possibilities of the Description Logic (quantifiers or types of relations). Their implementation in cartography corresponds with methods discussed in previous paragraphs. The main advantage of the approach using an application ontology consists in a homogeneous interconnection of data and semantics.</p><p> The real implementation of ontologies, other semantic resources and Linked Open Data principles in cartography can make web mapping development process more efficient, because the normalised semantic description enables to automatize many activities, including a derivation of new data and knowledge or checking of data as well as cartographic processes. Such an approach can bring the cartography closer to knowledge bases and systems and realise ideas of real-time cartography.</p><p> The research reported in this paper has been supported by the following project &amp;ndash; Sustainability support of the centre NTIS &amp;ndash; New Technologies for the Information Society, LO1506, Czech Ministry of Education, Youth and Sports.</p>

An Integrated Approach for Modeling Ontology-Based Task Knowledge on an Incident Command System

This paper presents the TTIPP methodology, an integration of task analysis, task ontology, integration definition function modeling (IDEF0), Petri net, and Petri net mark language (PNML), to organize and model the task knowledge in the form of natural language expressions acquired during the knowledge-acquisition process. The goal of the methodology is to make the tasks more useful, accessible, and sharable through the web for a variety of stakeholders interested in solving a problem which is expressed mostly in linguistic form, and to shed light on the nature of problem-solving knowledge. This study provides a core epistemology for the knowledge engineer while developing the task ontology for a generic task. The proposed model overcomes the drawbacks of IDEF0, which are its static nature and Petri net which has no concept of hierarchy. A good number of countries lie on the typhoon and earthquake belts, which make them vulnerable to natural calamities. However, a practical incident command system (ICS) that provides a common framework to allow emergency responders of different backgrounds to work together effectively for standardized, on-the-scene, incident management has yet to be developed. There is a strong need to explicitly share, copy, and reuse the existing problem-solving knowledge in a complex ICS. As an example, the TTIPP model is applied to the task of emergency response for debris-flow during a typhoon as a part of an ICS.