MONDIS Knowledge-based System: Application of Semantic Web Technologies to Built Heritage Riccardo Cacciotti 1 ; Jaroslav Valach 1 ; Martin Černansky 1 ; Petr Kuneš 1 1 Institute of Theoretical and Applied Mechanics AS CR, v.v.i, Prosecka 809/76, 190 00, Praha 9 1. Semantic web and cultural heritage As it is widely acknowledged, cultural heritage represents a complex domain characterised by a multidisciplinary approach to its documentation, management and protection. The increasing awareness concerning the importance of information sharing among practitioners worldwide advocates converging the heterogeneous perspectives towards a common understanding of concepts and mechanisms governing this research field. The main challenges posed to semantic web engineers derive from the fact that the data available in the domain usually present multi-format, multi-topical, multi-lingual, multi-cultural and multi-targeted features [Hyvoenen, 2012, 4]. Nevertheless semantic web technologies offer an important contribution to artificial data processing and they can positively affect pursuing the main scopes of built heritage conservation. In the context of computer and information sciences, an ontology defines a set of representational primitives with which to model a domain of knowledge or discourse. The representational primitives are typically classes (or sets), attributes (or properties), and relationships (or relations among class members). The defi nitions of the representational primitives include information about their meaning and constraints on their logically consistent application. Due to their independence from lower level data models, ontologies are used for integrating heterogeneous databases, enabling interoperability among disparate systems, and specifying interfaces to independent, knowledge-based services [Gruber, 2009]. In recent years there has been a considerable effort in exploring the application of ontologies to cultural heritage and promising findings have been revealed: indeed ontologies successfully enhance a systematic but fl exible organization of expert knowledge, describing comprehensively the interrelations and complementariness among different factors involved in the understanding of diagnostic and design problems. 2. MONDIS project 1 The on-going project has the objective to create a knowledge-based system able to replicate in a computer readable form the basic dependence among factors infl uencing the description, diagnosis and intervention of damages to immovable cultural heritage objects. The outputs include the launching of a multilingual, web-based interface and the development of context-specific applications. 2.1. System 2.1.1. Model description The MONDIS system exploits an ontological representation of the cultural heritage domain for the purpose of knowledge mapping and sharing conveying enhanced user accessibility and reliability of contents. The Monument Dama- 943
ge Ontology [Blasko et Al., 2012] is shown in figure 1. The cultural heritage domain is represented by concepts, referred to as classes, related by semantic connections. The model can be subdivided into thematic clusters: Cluster a) It refers to object and component identification allowing the documentation of general information such as construction type (e.g. tower, house, fountain), structural type (e.g. vertical cantilever, frame, load-bearing walls), functional type (e.g. barn, museum, church), use and style as well as the identification of component type (wall, floor, roof) and its related material(s). Cluster b) It refers to events, that is, occurrences which affect the conditions of the object. Event class is characterised by a temporal reference (either precise date or time range of occurrence) and includes location characteristics change (hydrogeological, geomorphological, weather) natural disasters (earthquake, fl ood), functional changes (object formally used as church is reused as restaurant), structural changes (component addition, removal or replacement), manifestation of damage (crack, deformation, collapse) and intervention (strengthening, cleaning, rehabilitating). Fig.1 - The Monument Damage Ontology 944
Cluster c) It concerns damage description, diagnosis and intervention, including the identifi cation of the typology of the damage detected on the object, referred to as Manifestation of Damage class, (e.g. component or material damages), the determination of the causing mechanism (e.g. bending, capillary rise, expansion) and of the carrying factor, the agent, which performs the mechanism (e.g. force, water, temperature). Manifestation of damage, mechanism and agent classes are connected to the intervention class, which encompasses all those actions aimed at repairing the MoD, stopping the mechanism and removing the agent. Cluster d) It refers to risk assessment. This cluster relates the hazard at a location with the vulnerability and value of a component or object, thus endorsing the verification of its risk with respect to the happening of certain disaster and consequently the evaluation of whether to employ or not retrofitting actions. Cluster e) It is dedicated to measurement ontology. Measurable entities in the model, such as MoD, agent, risk, component, material, can be assigned with measures of their quantities (e.g. width of crack, Young s modulus of a brick, height of a wall) and qualities (e.g. vulnerability level). The presented ontology is also enriched by the use of taxonomies, which consist into structured classifi cations of each of the concepts formalised into the model. Such taxonomies can be derived from relevant literature or can also be integrated or extended by the input of the community of users of the system. An example of the fi rst case might be the classification of stone degradation, prepared by ICOMOS in a glossary form, which has been easily integrated into the Manifestation of Damage class of MONDIS ontology [Snethlage, 2010]. 2.1.2. Functions MONDIS system supports the digitalization of a wide range of data including professional reports, books, articles and scientific papers. Differently from conventional databases, MONDIS is also able to provide automatic reasoning supporting and facilitating user s interaction and retrieving more efficiently relevant information from stored data. Furthermore the system is characterised by its capacity to continuously extend its contents and update its reasoning schemes accordingly with the level of knowledge stored by reporters. Two main functions of the system are considered, namely inputting and searching. Input follows model patterns in a way that resembles professional reasoning. Two typologies of entries can be distinguished: general knowledge entries and case study entries. The fi rst one refers to generalised rules for connecting two concepts in the model. These rules can be found in books, paper and articles: e.g. the information masonry has very low tensile strength provides a general rule for connecting masonry material to low tensile strength qualitative measurement. Its generality derives from the fact that such statement can be considered valid for all masonries. A practical example of inputting general knowledge in the system can be found in the terminology editor in which terms used in the ontology can be defined, translated and pictures can be attached (fi g.2). Case study entries, on the other hand, provide exact rules that cannot be generalised. Case studies present in fact peculiar rules for relating 945
Fig.2 - Entry example from terminology editor Fig.3 - OntoMinD visualization tool concepts within the model. An example can be Prague Castle has 2mm wide cracks on the external side of the walls due to freeze and thaw cycles. The statement crack has 2mm width is not in fact true for all cracks and therefore is specifi c to the case is being inputted. Throughout the inputting function, the user is guided through the model by the artificial intelligence constituted by the formalised relationships in the model and also by previously stored information. The system allows processing of incomplete information (i.e. no minimum input is required). Searching the system performs a computer-aided retrieval of information tailored to user s level of knowledge. Considering searching for a general manifestation of damage biological colonization, the system would provide to the user the stored information concerning for example the components on which this damage can be found, the mechanism that has generated it, possible repairs etc. Whenever the search if refined, by setting further parameters such as insect infestation of wooden beam, the knowledge retrieved narrows down to a smaller set of data with higher granularity (i.e. with deeper level of knowledge). As for inputting, searching function is also supported by automatic reasoning that direct user s query towards meaningful data. 946
Currently an OntoMinD based tool [Al-Jadir et Al., 2010] is used for the visualization of the ontological model and for testing of its functions (fig.3). 2.2 Applications to built heritage conservation MONDIS addresses different groups of users, namely administrators, professionals and researchers. The main form of interaction with the system is represented by a web-based user interface. However context specific applications are currently being developed. During the on-going phase of the project two applications are under testing and a user community is being created. The applications are: 1) MONDIS Mobile Application. Application for mobile devices aimed at gathering and processing information related to in-situ surveying (damage inspection and monitoring). 2) MONDIS Knowledge Matrix. Easy-learning tool for non-expert users, such as students and owners of historic buildings. Applications run on the ontological model by exploiting an inference engine. Such engine allows taking ontological knowledge, it deduces new knowledge and finally it provides the information back to the application user. The new inferred knowledge helps the application user to reveal new dependencies in the domain, or in other words to obtain coherent answers to the queries posed by the application itself. 3. Conclusions The research concentrates on the establishment of a knowledge-based system able to document and process the basic dependences among factors influencing the damages to immovable cultural heritage objects. The MONDIS system exploits the benefi ts of the latest sematic web technologies by formalising a comprehensive ontological representation of the cultural heritage domain. Two main functions are considered, namely inputting and searching. It is foreseen also the development of a number of context-specific applications, two of which are currently under testing. This work has been supported by the grant No. DF11P01OVV002 Defects in immovable cultural heritage objects: a knowledge-based system for analysis, intervention planning and prevention of the Ministry of Culture of the Czech Republic. Notes 1 www.mondis.cz References Blaško M., Cacciotti R., Křemen P., Kouba Z., 2012, Monument Damage Ontology, Lecture Notes, «Computer Science», 7616, 221-230. Gruber T., 2008, Ontology, Encyclopedia of Database Systems, Ling Liu and M. Tamer Özsu (Eds.), 1963-1965. Hyvoenen E., 2012, Publishing and Using Cultural Heritage Linked Data on the Semantic Web, Synthesis Lectures on the Semantic Web: Theory and Technology, 1-119. Snethlage R., 2010, Illustrated glossary on stone deterioration patterns, ICOMOS- ISCS. 947