An Ontology-Based Framework for Supporting Decision-Making in Conservation and Restoration Interventions for Cultural Heritage
Authors: 
Efthymia Moraitou, Yannis Christodoulou, Konstantinos Kotis, George CaridakisAuthors Info & Claims
Article No.: 41, Pages 1 - 24
Abstract
Decision-making (DM) is the backbone of the Conservation and Restoration (CnR) of Cultural Heritage (CH). The demands of the DM process for information organization and management have raised issues that the CnR community attempts to solve by creating DM-support tools and systems, which, among others, exploit Semantic Web (SW) technologies. Regarding the tools and systems that focus on the DM process of selecting an intervention option (CnR-DM-I), they present benefits, as well as limitations, regarding the (1) completeness of representation of the relevant knowledge in a unified manner, (2) facilitation of recording the CnR-DM-I process per se, in terms of the problem at hand as well as the intervention parameters, requirements, and criteria, and (3) recommendation and further exploration of CnR intervention options in a systematic manner. This work proposes an ontology-based framework as a means to overcome those limitations. The proposed framework (DS-CnRI) sets at its core a formal ontology which provides the necessary entities to represent expert knowledge related to CnR-DM-I. The ontology also includes rules which provide useful inferences to assist the CnR-DM-I process. The proposed framework has been deployed and evaluated in collaboration with conservators. Initial evaluation results show that the framework assists conservators in CnR-DM-I to detect and select the most suitable intervention options, to better understand the limitations of different options, and to document the process of reaching their decision.
1 Introduction
The Conservation and Restoration (CnR) domain aims to maintain the physical, aesthetic, and historical integrity of Cultural Heritage (CH) [1, 2]. In many cases, the intended insurance of tangible CH safeguarding is achieved through the implementation of CnR interventions (either preventive, remedial, or restorative) on conservation objects1 or their environment. To decide if and how to act regarding a particular CnR task, as well as to choose the most appropriate intervention method, conservators must go through a multi-stage, multi-factor Decision-Making (DM) process (from now on, Conservation and Restoration-Decision Making (CnR-DM) process) [4, 5]. The CnR-DM may be perceived as the backbone of the whole CnR process, since it possesses a central role in the day-to-day work of the conservator [4, 5]. Although it may differ from case to case, in general, the CnR-DM process essentially consists of up to six phases [1, 6]:
(1)
Initiation of the CnR project: Identification and specification of the problems, requirements, and objectives of the project based on the analysis, observations, measurements, and diagnosis conducted to understand the original, current, and desirable state of the conservation object at hand.
(2)
Identification of potential risks: Identification of potential risks of CnR interventions regarding the conservation object, health and safety risks for the conservators, and compatibility of the CnR interventions to be conducted.
(3)
Consideration of options and selection of CnR actions: Consideration over the suitability of different CnR intervention options, so as to meet the specifications recorded in previous stages.
(4)
Design of the CnR plan: Design of the CnR intervention plan agreed by all involved parts, including the conservators and other specialists such as curators, conservation scientists, and so on.
(5)
Implementation of the agreed plan: The actual CnR intervention (which may concern either the conservation object per se or its environment).
(6)
Completion of the CnR project: Recording and assessment of the implementation outcome, as well as of guidelines for future maintenance and handling of the conservation object or its environment.
During the CnR-DM process, the conservators collect and exchange various parts of diverse information and transform them into specific intervention options, and eventually into intervention decisions [5, 7]. Documentation2 and its management constitutes an integral part of CnR, as it occurs during all the different stages of the CnR-DM process [5]. The information that conservators need to take into account and combine to eventually choose the most suitable intervention may be relevant to (1) the characteristics of the conservation object at hand (materials, structure, damages, etc.), its environment (environmental parameters of the conservation object’s location), and any planned CnR intervention (i.e., any intervention that is considered to be applied on the conservation object), (2) CnR intervention options, including any techniques, supplies, and suitability requirements involved, (3) external factors, such as budget and location requirements/restrictions (e.g., power supply limitations), which must be taken into account so as to reach a decision [5, 6, 8, 9]. While all these pieces of information are crucial for assessing and deciding on actions, they are not always documented, or at least in a sufficient, consistent, and systematic way [8, 10].
To facilitate the management of the necessary information, and thereby support the CnR-DM process, several approaches have been proposed. Those include a wide range of solutions, from simple tools [10, 11, 12] to more sophisticated and complete systems [13, 14, 15]. Some of the tools and systems exploit Semantic Web (SW) technologies, especially in an attempt to tackle data interoperability (syntactic as well as semantic) and data exchange issues emerging during the process of accessing, comparing, combining, analyzing, and often visualizing CnR information—a task that is crucial for the overall support of CnR-DM [16].
The aforementioned tools and systems contribute to different stages and tasks of CnR-DM, while some of them particularly focus on choosing the most suitable intervention option for different cases at hand (from now on the DM process of choosing the suitable intervention will be referred to as CnR-DM-I). Those tools and systems present benefits as well as limitations regarding the (1) completeness of representation of the relevant knowledge in a unified manner, (2) facilitation of recording the CnR-DM-I process per se (i.e., how the expert reached a certain decision), in terms of the problem at hand and intervention parameters, requirements, and criteria, and (3) recommendation and further exploration of CnR intervention options in a systematic manner.
Drawing on the above, in this work, we propose an ontology-based framework (from now on, DS-CnRI), which aims to facilitate the DM process of CnR interventions, by assisting conservators to
—
organize the relevant information and determine (1) intrinsic requirements,3 which arise from the different intervention options and must be satisfied by the conservation object, its environment, or any planned interventions, (2) extrinsic requirements,4 which are defined by the conservator based on external factors and must be satisfied by the considered plan or its supplies, and (3) criteria,5 which refer to ranking criteria of suitable intervention options,
—
validate and enrich input information that is considered for the final decision (regarding the conservation object, its environment, and the planned CnR intervention), and
—
automatically receive a set of specific suitable as well as rejected intervention options for the case at hand, which can be further explored in terms of their characteristics and limitations.
The proposed framework sets at its core a formal ontology for the explicit representation and integration of expert knowledge related to CnR-DM-I. Particularly, the ontology provides all the necessary entities for representing specialized knowledge regarding the intervention problem, options, requirements, and criteria of two specific categories of CnR interventions: (1) the cleaning of superficial deposits6 from different surfaces, and (2) the consolidation of flaking gouache.7 Furthermore, it incorporates a set of rules which generate inferences regarding limitations of intervention options, based on the case at hand. The proposed framework consists of five stages:
—
Stage 1: Collection of information about the case at hand;
—
Stage 2: Identification of missing characteristics of the conservation object, its environment, and any planned interventions;
—
Stage 3: Identification and further assessment of both suitable and rejected intervention options;
—
Stage 4: Ranking of the suitable options; and
—
Stage 5: Collection of information regarding the final decision.
The rest of the article is structured as follows: Section 2 reviews semantic models that cover the domain of interest, as well as tools and systems that support CnR-DM, and discusses the fields that merit further research. Section 3 describes the architecture and workflow of the proposed framework and discusses its contribution to the CnR-DM process. Section 4 presents the underlying ontology together with the data management tools and processes. Section 5 presents the evaluation of the proposed framework which was conducted by professional conservators. Finally, Section 6 concludes the article with a brief discussion on obtained results and future research plans.
2 Related Work
Several tools and systems have been suggested for supporting CnR-DM. Michalski [10, 19] proposes the use of decision tables and decision trees for recording options, requirements, and criteria for the conservation of paintings. Similarly, Turk et al. [11] deploy decision tables for organizing criteria regarding the selection of suitable materials for the consolidation of stone, while in the context of DOCAM [12], decision trees are deployed for providing recommendations of strategies for the conservation of media art. Decision tables and trees could indeed be useful for the recording of different options, requirements, and criteria, for different cases at hand. Nevertheless, they do not provide interoperability and reuse of the recorded knowledge.
Complete Decision Support Systems (DSS) have also been proposed for assisting the CnR-DM process. Kupczak et al. [20] and Schalm et al. [21] have developed DSSs for identifying and predicting preservation problems of tangible CH through data analysis and visualization. In those cases, the systems contribute to the initial phases of the CnR-DM process, i.e., the specification of the problem and the evaluation of potential risk. In a different direction, the PRODOMEA DSS [13] was developed for facilitating the selection of suitable interventions by assessing the level of incompatibility of both past and planned interventions on masonry architectural assets. Moreover, La Russa and Santagati [15] propose an Artificial Intelligence-based DSS for suggesting the optimal combination of preventive conservation actions for museum collections through an algorithm that processes temperature and humidity measurements. Assessing the incompatibility level between interventions [13], as well as providing actual intervention suggestions [15], is considered significantly assistive. However, the underlying knowledge regarding the parameters for a certain case, the options, the requirements, and the criteria based on which a decision has been made, are not entirely captured in a unified, interoperable manner.
Regarding knowledge representation, it is important to mention that the CnR community has exploited SW technologies in an attempt to achieve data interoperability and exchange, which are both considered vital for the CnR-DM process. CnR lies within the wider CH domain, and therefore formal ontologies of the CH domain have been utilized for CnR data modeling. A popular example is the CIDOC CRM [22] and its compatible models [23] which include classes and relations that represent—to some extent— various CnR aspects [16, 24] and so they have been used for CnR data modeling [25, 26]. Another example is the more recent Architecture of Knowledge ontology network [27] which also covers some aspects of the CnR domain. In addition to using CH-related ontologies, the CnR community has developed specialized semantic models, which in some cases integrate and/or extend existing ontologies of the CH domain. Some examples are the Ontology of Paintings and Preservation of Art (OPPRA) [28], the Monument Damage Ontology (MDO) [29], the semantic model of the PARCOURS project [9], the ontology of the HERACLES project [14], the Conservation Process Model (CPM) [30], the Ontology for Degradation Phenomena and Annotation on 3D Reconstructions (ODPA-3RD) [31], the Conceptual Reference Model of French National Library (CRMBnF) [32], the Chinese Ancient Buildings Damages ontology (CABD) [33], and the Color and Space in Cultural Heritage Knowledge Representation (COSCHKR) [34]. Those models are specialized in different object types of movable and immovable CH (e.g., paintings, buildings), while they cover different CnR aspects of information (materials and technology, alteration, investigation, intervention, etc., for more see [16]).
The aforementioned models have been deployed in platforms and services which support different stages of CnR-DM. Particularly, OPPRA, MDO, PARCOURS, and HERACLES have been used in platforms and services for data integration and semantic search, reducing information retrieval time and improving quality of search results [9, 14, 28, 29]. Such services assist in problem identification, risk assessment, and retrieving intervention actions by providing unified access to the relevant information. Also, CPM and ODPA-3RD have been deployed in ontology-based visualization services which contribute to CnR-DM, by providing a more articulate and meaningful documentation as well as correlation of the requested information (e.g., the visualization of the extent and severity of an alteration phenomenon gives a thorough view of the conservation object’s condition) [30, 31]. Full-fledged ontology-based systems have also been proposed for supporting the CnR-DM. Zreik and Kedad [32] deployed the CRMBnF model into a DSS for identifying problems and prioritizing CnR interventions, through analyzing the conservation history of documents and enabling reliable predictions about their physical state. Additionally, Wang and Chen [33] exploit the CABD ontology for determining repair methods of Chinese buildings, by retrieving cases of damages and corresponding repair methods that present similarities with a given case. Finally, Boochs et al. [34] have deployed the COSCHKR ontology to a platform for assisting the selection of digitization and analysis methods of tangible CH cases.
Although the existing models may provide useful representations that cover some aspects of the CnR-DM-I process, they do not entirely cover the underlying knowledge in terms of the parameters, issues, requirements, criteria, intervention options involved, and their correlations [16]. Furthermore, the exploitation of this representation to provide services that will support the process of selecting valid intervention options can be further explored, since the existing approaches (1) facilitate the recording of the relevant information in a formal, structured, and unified way, but they do not support the recording and correlation of the different parameters, requirements, and criteria (e.g., [9, 14, 28, 29–32]) and (2) facilitate the discovery of CnR intervention, but they do not consider all the relevant requirements and criteria (e.g., [33]).
The bibliographic research showed that existing approaches support CnR-DM-I to some extent, providing the experts with useful information, general guidelines, or even specific intervention recommendations. Nevertheless, none of the proposed works (1) fully captures CnR-DM-I knowledge in a unified manner and (2) provides and explores suitable and rejected intervention options in a systematic manner (which will not solely rely on users’ searching), by taking into account different limitations and parameters for different cases at hand. Furthermore, the need for additional observation and understanding of the involved parameters at the stage of options consideration and selection is identified (especially in cases of semantic searching tools), but the explicit recording of the additional data which will enrich the documentation material of the conservation object is not supported [8, 10]. The proposed framework attempts to fill the aforementioned gaps of the existing tools and systems, exploiting the expressiveness of the ontologies for the formal representation of the experts’ knowledge.
3 The DS-CnRI Framework
3.1 Workflow
The DS-CnRI framework consists of five stages (Figure 1), described in detail below. To accommodate the description of the stages, we have established a use-case scenario according to which the conservator needs to decide how to clean the superficial deposits of the painting layer of a painting.
Fig. 1.
Stage 1: Information about the case at hand8 is collected. Stage 1 is further divided in four substages:
—
Substage 1.1: Information is collected based on CnR documentation about the conservation object, its environment, and any planned interventions. For instance, the conservator provides information about (1) the structural components of the painting (e.g., the painting has a painting layer, a substrate layer, and a support layer), (2) the materials of the components of the painting (e.g., the painting layer consists of acrylic paint), (3) characteristics observed on the components of the painting (e.g., the painting layer has flaking damage), and (4) the measured dimensions of the painting (e.g., the painting’s height is 2 m). Furthermore, the conservator may provide information about the environment of the painting, such as temperature, Relative Humidity (RH), and other environmental parameters. Additionally, the conservator may provide information about any planned interventions that will follow (e.g., wet cleaning will be applied on the conservation object after the intervention), as well as the supplies that the planned intervention involves (e.g., wet cleaning will use water at 40\({}^{\circ}\)C). Finally, information about the processes of measurement and observation (e.g., who and where conducted the measurement of conservation object’s height) can be collected for complete documentation.
—
—
Substage 1.3 (optional): Information about the extrinsic requirements is defined by the conservator, who can select among a number of extrinsic requirements involved in the issue at hand (e.g., absence of electric power).
—
Substage 1.4 (optional): The ranking criterion is specified by the conservator, who can select among a number of different criteria involved in the issue at hand (e.g., the performance speed of the suitable plan).
Stage 2: Missing characteristics of the conservation object, its environment, and any planned interventions, which are crucial for the final decision, are identified and collected. Stage 2 is further divided in two substages:
—
Substage 2.1: Characteristics are identified as missing in case (1) they constitute intrinsic requirements of considered options (e.g., the absence of flaking from the conservation object is required, a maximum (RH) 53% of the conservation object’s environment is required), and (2) relevant data have not been collected during Substage 1.1. At this substage, information collected during substages 1.1 and 1.2 (namely the information that describe the case at hand, and the issue that the CnR-DM-I is conducted about) and knowledge related to the considered intervention options (i.e., their intrinsic requirements) are exploited. For instance, Option A is considered as a solution for the cleaning of superficial deposits, and it has the Intrinsic Requirement A.1 which requires the absence of powdering11 from a painting layer. Since (1) the option is considered as a solution, (2) the conservation object at hand is a painting layer, and (3) there is no data regarding the absence or presence of any damage of such type on the painting layer (based on information of Substage 1.1), powdering is identified as a missing characteristic.
—
Substage 2.2: The identified missing characteristics of the conservation object, its environment, and planned interventions are collected (e.g., the conservator provides information regarding the absence of powdering damage on the painting layer).
Stage 3: The intervention options are identified as suitable or unsuitable (to be rejected) and are presented to the conservator for further assessment, by exploiting the collected data and the knowledge related to the intervention options. Stage 3 includes two substages:
—
Substage 3.1: The unsuitable (i.e., rejected) options are presented to the conservator. They are identified by exploiting the input data and the knowledge related to the intervention options. An intervention option is considered unsuitable (and therefore it is rejected) if it (1) is considered as solution for the case at hand, (2) has at least one intrinsic requirement that is not satisfied by the conservation object, its environment, or any planned interventions, and (3) involves plans and/or supplies that do not satisfy at least one extrinsic requirement. The conservator assesses the rejected options and the corresponding requirements along with the parameter that does not satisfy them. For instance, 32 options are considered for the problem of choosing a plan for the cleaning of superficial deposits. Twenty-six of the considered options have intrinsic requirements (Option A requires the absence of painting media from the conservation object, Option B requires the absence of flaking from the conservation object, and Option C requires the presence of paper on the adjacent layer which is substrate which are not satisfied by the painting layer (consists of acrylic paint and it has flaking though in local extent) and adjacent layers to the painting layer (the substrate layer is made of canvas rather than paper). Furthermore, 10 of the considered options involve plans requiring electric power (cleaning with a vacuum cleaner) and therefore they do not satisfy the extrinsic requirement of the absence of electric power use (the extrinsic requirement was selected in Substage 1.3 by the conservator).
—
Substage 3.2: The suitable options are presented to the conservator. They are identified by exploiting the input data and the knowledge related to the intervention options. An intervention option is considered suitable if it (1) is considered as solution for the problem at hand, (2) has intrinsic requirements that are satisfied by the conservation object, its environment, and any planned interventions, and (3) involves plans or/and supplies that satisfy the extrinsic requirements. For instance, 32 options are considered for the problem of choosing a plan for the cleaning of superficial deposits. Twenty-eight of those options (1) have at least one intrinsic requirement that is not satisfied by the painting layer and an adjacent layer to the painting layer, and/or (2) do not satisfy the extrinsic requirement of the absence of electric power use defined by the conservator. Therefore, only four of the options are suitable for the case at hand. The conservator assesses the suitable options including the involved plan (e.g., a plan involves mechanical cleaning technique and is applied locally on the surface to be cleaned), as well as the involved supplies (e.g., a plan requires the use of a soft brush).
Stage 4 (optional): The suitable options are ranked, based on the collected criterion set in Substage 1.4, and they are presented to the conservator. The ranking criteria refer to properties of the plans and/or supplies involved that correspond to ranking indexes. For instance, the ranking criterion of performance speed of the plan is selected and the four suitable options are ranked based on this criterion.
Stage 5: The final decision is made by the conservator and it is collected. The final decision can involve one or more suitable options. For instance, the final decision indicates Option A, which is one of the suitable options with the faster performance speed.
3.2 Framework Architecture
The technical implementation of the DS-CnRI framework consists of a number of components, which interact with each other forming a system. Below, these components and a high-level architecture of that system are presented (Figure 2):
—
ontology and rules (Component 1), which captures the expert’s knowledge related to CnR-DM-I. The ontology and the incorporated rules constitute the core of the framework and they provide all the assertions and inferences in the different stages of the framework.
—
data (Component 2), which include (1) the predefined data and (2) the collected data provided by the conservator about a case at hand. The data are modeled and organized based on the ontology.
—
reasoner (Component 3), i.e., the reasoning engine that processes the asserted data of the case at hand and produces the inferences.
—
query engine (Component 4), which enables the formation and execution of queries to get the corresponding retrieved data. The query engine must involve both asserted and inferred knowledge.
—
input data forms (Component 5), i.e., forms used for the collection of input data.
—
output data forms (Component 6), i.e., forms used for the presentation of retrieved data.
—
input data annotation (Component 7), which refers to semi-automatic mapping of the input data to the ontology.
The first four components constitute the knowledge base of the framework, which includes all the necessary knowledge/data and the tools for data management, including storage, inference, and retrieval.
Fig. 2.
Below, the interaction of the system components is described for each stage, according to the framework workflow (see Section 3.1):
—
In Stage 1, input data forms are used for the collection of necessary data for the case at hand. The collected data are then annotated to the ontology and stored in the knowledge base in the form of data.
—
In Stage 2, the reasoner processes the ontology and rules, as well as the data and produces inferences. The query engine is used to form and execute queries, exploiting the asserted knowledge (ontology, rules, and data) and inferences produced by the reasoner. During the same stage, the missing characteristics are presented through the output data forms. The missing characteristics that must be imported in the knowledge base to enrich the documentation are also collected using the input data forms, are annotated to the ontology, and stored in the knowledge base in the form of data.
—
In Stage 3 and Stage 4, the reasoner produces inferences based on the ontology and rules, as well as the data. Also, the query engine is used to form and execute queries. Finally, the retrieved data regarding the intervention options (unsuitable, suitable, ranked) are presented through output data forms.
—
In Stage 5, the suitable intervention options are presented in input data forms. One or more of them are selected, are annotated to the ontology, and stored in the knowledge base in the form of data.
3.3 Contribution to the CnR-DM-I Process
As mentioned in Section 1, although the CnR-DM process may differ from case to case, it generally includes the following tasks: (1) specification of the problem, requirements, and criteria, (2) potential risks assessment, (3) selection of the appropriate intervention, (4) design of the intervention, (5) implementation of the intervention, and (6) assessment of the outcome and recording of guidelines. The DS-CnRI framework supports the conservator’s workflow by facilitating the first four tasks, as follows:
—
Stage 1 facilitates the recording and organization of information about the conservation object, its environment, and any planned interventions, as well as the specification and documentation of the issue at hand, the extrinsic requirements, and the ranking criteria. In this stage, the aforementioned information is collected and mapped to the ontology, to be processed by the reasoner and the query engine in later stages. As a result, the CnR documentation data for the basic parameters, as well as important information that is often neglected, such as the issue of the CnR-DM-I process, the extrinsic requirements that the conservator may set, and the criteria based on which the suitable options will be ranked, are recorded and modeled during this stage in a systematic and unified way.
—
Stage 2 facilitates the recording of any extra information which is crucial for the effective assessment of intervention options, by encouraging the conservator to collect additional information about the characteristics of the conservation object, its environment, and any planned interventions. This stage may trigger the meticulous examination of specific characteristics which must be taken into account for the final decision. Stage 2 assists the expert to avoid possible risks due to incompatibility of options, as well as to repeat observations and recording in later stages (i.e., the conservator may realize that a certain characteristic which is important for the assessment of an intervention option has not been taken into account).
—
Stage 3 supports the assessment of intervention options and the identification of the suitable ones. Human reasoning regarding the assessment of the various intervention options can be proved very complex due to the potentially large number of the different parameters and requirements, and their interplay. In this stage, all the necessary reasoning is automatically performed, unburdening the conservator of this tedious task, so that they can better organize their thoughts and realize the requirements and potential limitations and risks in advance. Furthermore, the conservator can review and keep records of all the possibilities and restrictions for the case at hand in a systematic way, avoiding multiple and possibly inconsistent searches over the different options for a given issue.
—
Stage 4 supports the selection of the most suitable option based on certain ranking criteria. The conservator can easily rank and compare the suitable options in a systematic manner and choose the most preferred one according to different criteria.
—
Stage 5 facilitates the recording of the final decision as well as the design of the intervention implementation, since the decision is accompanied by a fully recorded plan describing its characteristics, as well as any supplies involved.
4 Implementation
4.1 The Ontology
For the development of the ontology, the Human-Centered Collaborative Ontology Engineering Methodology [36] was followed due to its collaborative, iterative, and human-centered nature. Each phase of the methodology was accompanied by four structured meetings with conservators from the National Museum of Contemporary Art Athens (EMST) and the National Gallery Alexandros Soutsos Museum (NGASM), as well as by asynchronous communication in the form of notes and comments on the shared documents of the team’s stakeholders, i.e., domain experts and the knowledge engineer.
During the conceptualization stage, we studied examples of CnR documentation data provided by the two conservation laboratories, based on condition reports and conservation process reports.12 The data included descriptions of conservation objects, their environments, and any proposed or implemented CnR interventions. However, data explaining the exact restrictions and criteria based on which the conservators selected a certain intervention plan were either absent or documented in free text, in an unstructured manner (e.g., in the description of the implemented intervention). As such, we engaged in a process of collecting and organizing such data, taking as an indicative example two categories of CnR interventions (1) the cleaning of superficial deposits and (2) the consolidation of flaking gouache. The data were collected based on (1) bibliographic research and (2) experts’ related knowledge and experiences (the data collected in Google Sheets [37], Google Docs [38]). The cleaning of superficial deposits is a very common intervention which is applied in a variety of different surfaces of conservation objects (regarding their materials, physical features, and general structure) for preventive conservation or as a first action of a remedial conservation project. On the other hand, the consolidation of flaking gouache is a more specialized intervention, applied only on the painting layers of artworks that are made with the gouache technique. With those two CnR intervention categories we were able to cover (1) both a more general and a more specialized case of intervention and (2) the different categories of requirements that have been identified based on our research. Particularly, the intrinsic requirements regarding the characteristics/dimensions of (1) the conservation object, (2) a physical feature of the conservation object, (3) an adjacent layer of the conservation object, (4) the environment of the conservation object, (5) a planned intervention for the conservation object, (6) the supplies involved in a planned intervention for the conservation object, as well as the extrinsic requirements and potential criteria regarding the characteristics/dimensions of (1) the considered plan for the case at hand and (2) the supplies involved in the considered plan for the case at hand.
At the same stage, existing semantic models were studied and analyzed. The models were selected based on the main concepts of the CnR-DM-I domain that we have extracted during data analysis. The study included (1) the research of the literature which was conducted using the data sources of Semantic Scholar, Springer Link, ScienceDirect, and AATA Online and searching for topics related to CnR, CH, Ontologies, SW, and CIDOC CRM, and (2) search in ontology repositories (LOV [39] and ODP [40]) for related terms such as conservation, DM, and intervention.
It should be mentioned that there were ontologies we considered relevant, though they were not reused in the developed ontology due to (1) availability limitations (e.g., usage and accessibility to Resource Description Framework versions of the models), (2) conceptualization differences (e.g., in our perception the criteria are related to the characteristics of the options and can prioritize equally suitable options, whereas in other conceptualizations the criteria express limitations that could exclude options as unsuitable), (3) introduction of axioms that were considered unsuitable for this approach (e.g., axioms for the correlation of options with criteria). We have also based our decision to reuse existing ontologies on criteria presented in the work of Kotis et al. [41], i.e., recent ontologies that are still ‘live’, are reused by or reuse other models. Nevertheless, some of those ontologies provided useful insights for our modeling decisions. For instance, CRMact [42] and CRMinf [43] include classes that are conceptually relevant, especially regarding the CIDOC CRM classes that they extend. Based on this observation, we have made a few design choices regarding the superclasses that the new added DCRI-ont (ontology for the DM in CnR interventions) classes extend. Also, we noticed that CRMinf expresses, at the data level, the functional part of our ontology (i.e. the rules), and therefore it could be used supplementarily to express the process of the reasoning achieved through the rules. Furthermore, the DM ontology [44] was not reused due to differences in some conceptualizations (regarding the requirements and criteria representation), although it was studied in terms of the design patterns proposed which we adopted in many cases. Eventually, the models that were thoroughly studied and chosen for reuse are (1) CIDOC CRM,13 (2) its compatible model CRMsci14 [45], (3) the CIDOC CRM extension about typed properties and negative typed properties15 [46, 47], and (4) the Simple Knowledge Organization System (SKOS)16 ontology [48].
Based on the communication with the experts and the data that they provided, several competency questions (CQ) were shaped collaboratively. A representative list of the CQs and corresponding answers are presented in Table 1. While some of the questions are more general and relatively simple (e.g., CQs 1, 2, and 3), some others are more specific, focusing on CnR-DM-I and therefore more complex (e.g., CQs 4, 5, an 6). We must highlight that the CQs have been also used as part of the retrievals for the framework that the ontology supports.
Table 1.
| CQs | Expected answer |
|---|---|
| CQ1: What are the materials of the Painting layer of Artwork 1000 (conservation object)? | Acrylic paint |
| CQ2: What options are considered for the issue of choosing a cleaning superficial deposits plan? | Option A, Option B, Option C |
| CQ3: What are the adjacent layers (if any) of the Painting layer of Artwork 1000 (conservation object)? | Support layer of Artwork 1000, Substrate layer of Artwork 1000 |
| CQ4: What are the features of the Painting layer of Artwork 1000 (conservation object) about which an intrinsic requirement requires their absence/presence and there is no relation of their presence/absence? | Relief, stickiness |
| CQ5: Which of the considered options, regarding the DM process 1 for the Painting layer of Artwork 1000 (DM of particular conservation object), have no unsatisfied intrinsic requirements? | Option A, Option B |
| CQ6: What is the order of the suitable options according to the criterion defined? | 1-Option A—performance speed fast 2-Option B—performance speed slow |
The ontology was developed in Protégé 5.5.0 [49] and it consists of three modules:
(1)
DCRI-ont, which includes all the necessary classes and properties for the representation of the CnR-DM-I process and also the defined rules. DCRI-ont directly imports CIDOC CRM, CRMsci, the CIDOC CRM extension about typed properties and negative typed properties, and SKOS.
(2)
DCRI-voc, which includes individuals that express types of different basic concepts of the CnR-DM-I domain (e.g., types of materials, types of CnR interventions, types of damages). DCRI-voc directly imports SKOS and, although it has been developed as part of the ontology, it can also be used independently as a SKOS vocabulary for the CnR-DM-I domain.
(3)
DCRI-ont-special, which includes classes, properties, and individuals required for the two selected intervention categories. It directly imports DCRI-ont and DCRI-voc (and by extension CIDOC CRM, CRMsci, CIDOC CRM extension about typed properties and negative typed properties, and SKOS).
The DCRI-ont includes core entities regarding the CnR-DM-I process which constitutes the base for its representation and can be further extended for different CnR intervention categories. On the other hand, DCRI-voc and DCRI-ont-special are more specialized and illustrate the capacity of the framework regarding the CnR-DM-I process. The working version of the ontology (including all the three modules) is available in OWL format via GitHub.17 The documentation of the ontology (detailed description of the three modules, scope notes of different entities, etc.) has been developed with WIDOCO [50] and can be accessed on the corresponding webpages (DCRI-ont documentation,18 DCRI-voc documentation,19 DCRI-ont-special documentation.20)
The classes and relations of the three ontology modules (DCRI-ont, DCRI-voc, and DCRI-ont-special) cover four distinct (though highly interlinked) thematic clusters:
(1)
CnR-DM-I process, which refers to the DM behind a CnR intervention conducted by a conservator. It includes entities that represent the decision-maker, the issue, the options, the requirements, and the criteria involved in the CnR-DM-I process. Additionally, it includes the necessary properties to represent the interconnections between the CnR-DM-I process and the considered parameters (e.g., the plans that the options involve, the parameter that does not satisfy a requirement, the conservation object that motivated a CnR-DM-I process).
(2)
Conservation object, which refers to the material and immaterial characteristics of the tangible CH. It includes classes and properties that represent administrative information (identification, ownership, preservation, and management), materials and technology (production materials and techniques, structural layers and components, qualitative characteristics), and alteration (deterioration) of the conservation object.
(3)
Conservation object’s environment, which refers to the environment of the conservation object. It includes classes and properties that represent quantitative and qualitative characteristics of the conditions of the conservation object’s location.
(4)
CnR intervention plans, which refers to planned actions that can be applied to the conservation object or its environment. They can be either general plans that can be applied to any conservation object/environment or specific plans that are designed for certain conservation objects/environments. It includes entities and relations that represent the plans, their aims, techniques, and supplies.
The main concepts and relations of the aforementioned thematic clusters are presented in Figure 3.
Fig. 3.
Regarding the reuse of CIDOC CRM and CRMsci classes and properties, it was conducted in two ways, depending on whether the concept to be represented constituted a specialization of or was semantically equivalent with CIDOC CRM/CRMsci classes and properties. In the first case, a new class/property was defined as a subclass/subproperty of some CIDOC CRM class/property. The classes that extend a CIDOC CRM class are presented along with their scope note in Appendix. Regarding the new properties, most of them do not extend a CIDOC CRM/CRMsci properties. There are 18 object properties that extend cidoc-crm:P2\(\_\)has\(\_\)type and cidoc-crm:P2i\(\_\)is\(\_\)type\(\_\)of respectively, and 5 data properties that extend cidoc-crm:P90\(\_\)has\(\_\)value. These properties specialize CIDOC CRM properties to accommodate the correlation of DCRI-ont classes (for a more detailed presentation of the object properties see documentation sites21,22). In the second case, the equivalent CIDOC CRM/CRMsci class has been identified and marked for future data modeling. Furthermore, a number of proposed typed and negative typed properties were imported from the respective CIDOC CRM extension to correlate individuals of parameters with individuals of types based on their existence/absence. Overall, 48 classes (37 in DCR-ont and 11 in DCRI-ont-special), 77 object properties (55 in DCR-ont and 22 in DCRI-ont-special), and 10 data properties (4 in DCR-ont and 6 in DCRI-ont-special) were created. Most of the novel classes and properties focus on the Thematic Clusters 1 and 4.
Apart from the classes and properties, the ontology includes several individuals which capture specialized knowledge regarding (1) the cleaning of superficial deposits and (2) the consolidation of flaking gouache. Those individuals are related to (1) specific types of different basic ontology concepts, (2) qualitative values, (3) issues, (4) CnR intervention plans, (5) CnR intervention options, (6) requirements, and (7) criteria. The individuals representing the specific types of concepts are members of (1) the SKOS class Concept and (2) some sub-class of the CIDOC CRM class E55\(\_\)Type. In cases where a term is narrower or broader, then the respective individuals are interrelated through the SKOS has\(\_\)narrower/has\(\_\)broader properties. On the other hand, the individuals representing the qualitative values, issues, plans, options, requirements, and criteria are members of some class of the proposed ontology. Finally, all collected data provided by the conservator about a case at hand are stored in the ontology as ABox individuals.
Furthermore, some additional knowledge must be inferred, and a set of rules has been defined in the form of “IF-THEN” using SWRL [51] and the Protégé SWRLTab plugin [52]. The defined rules produce five types of inferences (Rules Categories 1–5):
(1)
The intervention options that must be considered in the context of a CnR-DM-I process (RC 1). The options are directly correlated with the issues they resolve. The CnR-DM-I process is directly correlated with the issue at hand. Based on these asserted relations, the relation between the CnR-DM-I process and the options to be considered is inferred (Figure 4, object property dcri-ont:considers).
(2)
The intrinsic requirements in the context of a CnR-DM-I process (RC 2). The intervention options are directly correlated with their intrinsic requirements, while the intrinsic requirements are directly correlated with the types of parameters for which they must be taken into account. On the other hand, the CnR-DM-I process is directly correlated with the conservation object at hand. Furthermore, the conservation object is correlated with its environment and any planned activity, all of which have a particular type. Based on these asserted relations, the intrinsic requirements of the CnR-DM-I process are inferred (Figure 4, object property dcri-ont:stipulates).
(3)
The satisfaction of requirements (both intrinsic and extrinsic) regarding the absence/presence of characteristics (RC 3). The characteristic (e.g., a physical feature such as powdering, a qualitative value such as slow performance speed), as well as the type of the parameter (e.g., structural layer) that must satisfy the requirement, is directly correlated with the requirement. On the other hand, the characteristic or qualitative value that a parameter has and the type of the parameter are directly correlated with the parameter. Based on these asserted relations, the parameters that do not satisfy the requirements are inferred (Figure 4, object property dcri-ont:isNotSatisfiedBy).
(4)
The satisfaction of requirements (both intrinsic and extrinsic) regarding the threshold value of a dimension of a parameter (RC 4). The dimension type (e.g., RH), the threshold value (e.g., maximum 50%), as well as the parameter type (e.g., exhibition environment) are directly correlated with the requirement. On the other hand, the parameter is correlated with a dimension which has a dimension type and a value. Based on these asserted relations, the relation between the parameter and the corresponding unsatisfied requirement is inferred. It is very similar to the previous category but this time values of dimensions—such as height or temperature—are taken into account for the inference.
(5)
The relations of terms which describe different CnR-DM-I parameters and broader/narrower terms of the CnR domain (RC 5). The broader/narrower relations are directly correlated with different terms (e.g., the term Canvas is narrower than the term Textile). The terms are directly correlated with different parameters (e.g., the substrate layer consists of Canvas). Based on these asserted relations, the indirect relation between the parameters and terms of broader/narrower meaning is inferred. For instance, based on the assertion that the substrate layer consists of Canvas it can be inferred that it also consists of Textile (Figure 4, object property cidoc-crm: P45\(\_\)consists\(\_\)of).
Fig. 4.
4.2 Data
As we already mentioned in Section 4.1, the ontology and the rules have been developed in Protégé 5.5. The same software has been used for the storage of collected data provided by the conservator about a case at hand. The data are stored in the form of individuals based on the ontology schema.
4.3 Reasoner
The Pellet reasoner [53] has been used for producing all the inferences based on the ontology and the rules, as well as the input data for the case at hand. Pellet can find any inconsistency of the ontology, compute the classification hierarchy, and most importantly provide and explain inferences. The inferences are particularly important for the framework, since they are exploited for producing the final outputs presented to the conservators at Stages 3 and 4, regarding (1) the requirements that are not satisfied together with the parameter that does not satisfy them, (2) the suitable options, (3) the rejected options. Figure 5 presents an example of three rules (in natural language) together with the inferences they produce. The asserted relations involved in the reasoning are presented in Section 4.1, Figure 4. We must highlight that the end user, the conservator, indirectly exploits the inferences through the results of the output forms.
Fig. 5.
4.4 Query Engine
In the context of the framework, a number of queries must be executed taking into account both asserted and inferred knowledge. These queries represent the CQs that were defined for ontology development (see Section 4.1). The queries are formulated in SPARQL language [54], and they are grouped in accordance to the different substages that involve retrievals (namely Substages 2.1, 3.1, 3.2, and 4). The order and the need of the queries execution for each substage is standardized and can be applied for different cases at hand. For instance, in Substage 2.1, a query will be executed (see CQ4, in Section 4.1) to retrieve types of physical features in which absence/presence is required by one or more of the considered options, but the conservator has not provided any relevant information in Substage 1.1 (see example in Figure 8). Another example is a query which is executed in Substage 4 (see CQ6 in Section 4.1) to retrieve and rank the suitable options (see example in Figure 7). The queries are executed through the Snap SPARQL plugin [55]. It should be noted that the end user, the conservator, does not interact with the query engine but only with the retrievals presented in the output forms.
4.5 Input/Output Data Forms
Regarding the development of forms for input/output data (in other words for the collection and presentation of data), MS Excel files have been used. The necessary forms have been formulated according to the different substages of the framework’s workflow.
The input data forms provide fields for data input which are either simple text boxes or drop-down lists. For instance, data about the issue at hand and the materials that the conservation object consists of are input via drop-down lists, while the identification number of a conservation object or the date of the CnR-DM-I process are input via simple text boxes. An example of input data form is presented in Figure 6.
Fig. 6.
On the other hand, the output data forms provide tables for organizing the query results. We must highlight that in the stage of detecting missing characteristics (Stage 2) we have a combination of output/input data, therefore the corresponding form contains both input and output elements. There is a fill-in layout where the user selects a statement of absence or existence from a drop-down list for a particular characteristic that has been retrieved as missing. Examples of the two cases of output data forms are presented in Figures 7 and 8.
Fig. 7.
Fig. 8.
4.6 Annotation of Data
The annotation of collected data provided by the conservators about the case at hand is conducted in a semi-automatic manner using Cellfie plugin [56] and the MappingMaster [57], a domain-specific language that Cellfie uses for importing data from MS Excel files to Protégé. Before mapping and importing the individuals in the ontology, the collected data are pre-processed to apply the necessary predefined syntax, using MS Excel functions (e.g., copy, replace, conjunction). Next, the data, which are organized in separate thematic tabs (e.g., data for conservation object description, data for measurement process description), are imported in Protégé using the Cellfie plugin and exploiting a set of already formulated mapping rules (e.g., see Figure 9). The template of the MS Excels for the organization of the data as well as the mapping rules can be reused for different cases. We have to highlight that the conservator is not involved in pre-processing and mapping of the data but only in the collection of the data in the MS Excel forms.
Fig. 9.
5 Evaluation
The evaluation of the proposed framework has been conducted following a user-centered qualitative approach. Eight conservators participated in the evaluation. Six of the conservators have a postgraduate education level, while one of them has a PhD education level. All of them work in different conservation laboratories of (1) museums (EMST and NGASM), (2) research center (Hellenic Center for Research and Conservation of Archeological Textile-ARTEX), (3) conservation projects conducted by Directorates and Ephorates of the Greek Ministry of Culture. They also have different conservation specialization (in paintings, contemporary art, paper, book and archival material, textile, organic materials), while six of them have an experience of more than 10 years in the domain. Regarding the CnR documentation process, one of the conservators has a very high occupation with documentation process in their day-to-day work, while five of the conservators have high and two of them medium occupation, respectively. Additionally, two of the conservators use only MS Word, four use MS Excel, one uses MS Word and MS Excel, and one uses MS Word and an information system for CnR documentation. The evaluation was conducted in separate one-hour meetings.
Table 2.
| Axis of question | Number of question | Question |
|---|---|---|
| Axis a | Q1 | I would use the framework for the recording of more CnR intervention categories |
| Axis a | Q2 | Using the framework, it is possible to represent the potential requirements which may lead to the rejection of an option |
| Axis a | Q3 | Using the framework, it is possible to represent the potential criteria for the selection of an option |
| Axis a | Q4 | I would use the framework for the recording of CnR-DM-I processes |
| Axis b | Q5 | The framework assists me in the specification of the requirements and the criteria which may lead to the selection of an option |
| Axis b | Q6 | The framework assists the observation of characteristics of parameters that may influence the final decisions |
| Axis b | Q7 | The framework assists me in identifying extra requirements apart from the ones that I had already thought by myself |
| Axis b | Q8 | The framework assists in the better justification of the rejection of an option |
| Axis b | Q9 | I would use the framework to check the correctness of a selection |
| Axis c | Q10 | Using the framework, I found all the suitable options that I had already thought by myself |
| Axis c | Q11 | Using the framework I found extra suitable options regarding the ones that I had already thought by myself |
| Axis c | Q12 | The framework assists me in finding quickly the suitable options |
| Axis c | Q13 | The framework assists me in deciding quickly for the suitable option |
| Axis c | Q14 | I would use the framework to discover potential options for a case at hand |
At the first part of the evaluation, the conservators used the framework to make CnR intervention decisions about (1) the cleaning of superficial deposits and (2) the consolidation of flaking gouache, for different example-cases of conservation objects of museum collections. In this context, descriptions of conservation objects and their condition state were created. Furthermore, specifically for the consolidation of flaking gouache, data describing planned interventions and the environment of the conservation object were created. Additionally, data about the process of measuring and assessing the conservation object, its environment, and any planned interventions were created. The data were collected in an MS Excel template by the conservators, and subsequently they were pre-processed and annotated (Component 7) by the ontology engineer (Substage 1.1).
The conservators used the input/output data forms (Components 5 and 6) for (1) documenting the CnR-DM-I process, in terms of problem, date, decision-maker, extrinsic requirements, and ranking criterion (Substages 1.2–1.4), (2) presenting and inputting the existence/absence of missing characteristics (Substages 2.1–2.2), (3) presenting the rejected options, along with the unsatisfied requirements (Substage 3.1), (4) presenting information about the plans and supplies of suitable options (Substage 3.2), (5) presenting the ranking of the suitable options (Stage 4), and finally, (6) inputting the decision (Stage 5). The ontology engineer facilitated the (1) importing of user data from the MS Excel input forms to the knowledge base (executing MappingMaster rules through Cellfie, Component 7—Stage 1, Substage 2.2, Stage 5) and (2) presenting the retrieved data in the MS Excel forms (by executing SPARQL queries through SnapSPARQL and transcribing them in the forms, Component 4— Substage 2.1, Stages 3 and 4).
At the second part of the evaluation, the conservators were engaged in a discussion, regarding the overall experience of the use and performance of the proposed framework. The discussion was structured in three main axes: (a) the organization and recording of the information related to CnR-DM-I process, (b) the identification and realization of restrictions that influenced their decision, and (c) the discovery and selection of intervention options. Additionally, they answered a set of questions (see Table 2). Each question corresponds to a discussion axis. The answers were given in a Likert scale from 1 (completely disagree) to 5 (completely agree). Before answering the main questionnaire, the users answered some general questions regarding their background in the CnR domain and their experience in CnR documentation (presented at the beginning of Section 5).
The answers for each question are presented in Figure 10, showing an overall positive consideration of the framework and its usage. Regarding the first axis, the conservators agreed that the proposed framework indeed contributes to the complete recording of information related to the CnR-DM-I process, and that the expression of involved options, parameters, requirements and criteria is satisfying. Furthermore, seven of the respondents were very positive in using the framework for the recording of the CnR-DM-I process. Regarding the second axis, the conservators stated that the framework helped them in realizing the existence of requirements and thus limitations. Four of the respondents stated that this realization in some cases confirmed and even strengthened their final decision. Furthermore, they positively evaluated the input of missing data for the case at hand. In four cases, the framework assisted the conservators to realize that there were characteristics of the conservation objects that they had not included in their initial documentation, which were crucial for the selection/rejection of an option, as well as for documenting the condition state of the conservation object. Finally, regarding the third axis of the discussion, the conservators were satisfied by the assistance in discovering and selecting intervention options. They confirmed the options that they already had in mind, and seven of them agreed that they also had the chance to see additional options that they had not considered. Only one conservator stated that found solutions that already had in mind, and all the options had already been considered. Regarding the framework’s support in making the final decision, all the conservators were positive, and they agreed (four out of eight) or strongly agreed (four out of eight) that the framework assisted them in deciding quickly. Five of them also found that using the framework encouraged them to further assess their final decision (e.g., by exploring the ranking of the suitable options or by exploring the limitations of the rejected options) since all the possibilities and limitations were presented to them. In general, they were willing to use the framework in future tasks. Finally, all of them stated that it could be particularly useful in categories of CnR interventions that they are not very familiar with, while five of them stated that it could be useful in more demanding cases, that would require a more thorough planning and specific arguments for making final decisions.
Fig. 10.
6 Conclusions and Future Work
In this work, we proposed an ontology-based framework which aims to assist the conservators to (1) organize information and determine requirements and criteria over the CnR intervention at hand, (2) validate and enrich input information that is considered for the final decision, and (3) identify suitable and rejected intervention options for the case at hand, which can be further explored in terms of their characteristics and limitations. For that purpose, the framework exploits the expressiveness of formal ontologies for the representation of the underlying expert knowledge. The proposed framework has been deployed and evaluated based on certain case studies in collaboration with conservators from conservation laboratories in Greece.
The work so far indicates that the developed ontology efficiently represents the domain of interest and supports the objectives of the proposed framework. Two different categories of CnR interventions have been successfully modeled so far, including different categories of intrinsic and extrinsic requirements that impact the selection of suitable intervention options. The proposed framework has been positively evaluated by experts since it contributes to the complete and unified documentation of the CnR-DM-I process, and it successfully provides suitable CnR intervention options, supporting the day-to-day work of the professionals. The evaluation will be continued in order to collect more opinions from experts regarding its usage, effectiveness and potential.
Future work will focus on the utilization of the ontology and the framework beyond the existing case studies. More categories of CnR interventions will be added, enriching the knowledge base and improving the general usefulness of the framework for the conservators. For instance, categories of preventive conservation intervention will be included, while more sophisticated CnR interventions that are not very common will be added to more completely evaluate the usefulness and possible limitations of the proposal. Also, the framework will be tested against more real cases to spot and resolve any reasoning limitations as well as to obtain a quantitative analysis of the system’s performance. Furthermore, the development of a complete system that incorporates the framework and executes the different processes seamlessly, automating the data input/annotation processes where possible, would be particularly valuable for the end users and therefore is considered a primary future goal for the establishment of the framework. Finally, regarding the wider adoption of the proposed ontology, it would be useful to continue the study of the CIDOC CRM and its compatible models and engage in more thorough discussions with relevant working groups of the CIDOC CRM community. The discussions should focus on the observations and proposals of the research to verify its integration and field of contribution compared to the existing models.
Acknowledgments
We would like to express our gratitude and sincere appreciation to Foteini Alexopoulou, conservator of the National Museum of Contemporary Art Athens (EMST), and Elina Kavalieratou, conservator of the National Gallery Alexandros Soutsos Museum (NGASM), for their collaboration and the sharing of their expertise that contributed in crucial parts of the research. Additionally, we have to specially thank all the conservators that participated in the evaluation phase of the research.
Footnotes
1
Conservation object refers to the object that is “worthy of conservation, and not only repair, maintenance, cleaning, or care” [3].
2
3
For instance, an intervention option has an intrinsic requirement, based on which the absence of powdering from the conservation object is required.
4
For instance, an intervention option does not satisfy an extrinsic requirement, based on which the use of electric power for the execution of the intervention is forbidden.
5
For instance, a criterion for the decision between suitable options, is the performance speed of the intervention.
6
It refers to the reduction of superficial soil, dust, grime, insect droppings, accretions, or other surface deposits of conservation objects [17].
7
It refers to the stabilization of flaked areas of gouache painting layers by introducing materials [18].
8
The case at hand refers to the parameters, i.e., the conservation object, its environment, and any planned interventions, as well as the issue at hand, the decision-maker, the date, and any extrinsic requirements and criteria.
9
The decision-maker is the conservator.
10
In the particular case study, the CnR-DM-I concerns (1) a structural layer of a complex object or (2) a solid object, which may be either a part of a complex object or a distinct object.
11
Powdering refers to the act or process of reducing to powder, pulverization; in conservation science, it refers to granular disintegration of stone and pigments [35].
12
Condition report refers to the document that records the existing condition of the conservation object(s), in terms of preservation state, prior to any CnR intervention (e.g., treatment, moving) [8]. The report of the conservation process refers to the written report that records the methods, materials, and equipment used on a conservation object for the treatment of different undesirable characteristics [8].
13
https://cidoc-crm.org/sites/default/files/cidoc_crm_v6.2.1-2018April.rdfs (FORTH-ICS, version 6.2.1)
14
https://cidoc-crm.org/crmsci/sites/default/files/CRMsci_v1.2.6.rdfs (FORTH-ICS, version 1.2.6)
15
https://github.com/linked-conservation-data/crmntp (Linked Conservation Data consortium)
16
http://www.w3.org/2004/02/skos/core# (W3C-SKOS)
Appendix
| DCRI-ont class | Scope note | Super-class |
|---|---|---|
| ConservationAndRestoration Intervention Decision-maker | It refers to the conservator which performs the decision-making process to decide for a conservation and restoration intervention. | cidoc-crm:E39_Actor |
| ConservationAndRestoration InterventionSupply | It refers to the supply that an intervention activity employs or an intervention procedure specifies. | cidoc-crm:E22_Man-made_Object |
| CommercialAppellation | It refers to the commercial appellation that an intervention supply may have. | cidoc-crm:E41_Appellation |
| ConservationAndRestoration InterventionPlan | It refers to a plan (or a standard process) which aims to the preservation or/and the restoration of the object to the best possible condition, through the direct or indirect (i.e., environment) intervention to a conservation object (of any kind, movable or immovable CH). | cidoc-crm: E29_Design_or_Procedure |
| ConservationAndRestoration Intervention Decision-makingCriterion | It refers to the criterion that a decision-making process stipulates, regarding the characteristics of a plan or/and the involved supplies. | cidoc-crm:E89_Propositional_Object |
| ConservationAndRestoration Intervention Decision-makingOption | It refers to the option that a decision-making process considers as possible to be chosen for a certain issue. | cidoc-crm:E89_Propositional_Object |
| ConservationAndRestoration Intervention Decision-making Requirement | It refers to the requirements that are stipulated by a decision-making process. It may be a requirement of an option (intrinsic requirement) or a requirement based on an external influence factor (extrinsic requirement) (e.g., the budget of a conservation project). | cidoc-crm:E89_Propositional_Object |
| ActivityType | It refers to the type of an activity. | cidoc-crm:E55_Type |
| ActorType | It refers to the type of an actor. | cidoc-crm:E55_Type |
| ConservationObjectType | It refers to the type of a conservation object (a whole object or a part of it). | cidoc-crm:E55_Type |
| ConditionStateType | It refers to the type of a condition state. | cidoc-crm:E55_Type |
| DocumentType | It refers to the type of a document. | cidoc-crm:E55_Type |
| EquipmentType | It refers to the type of equipment. | cidoc-crm:E55_Type |
| PhysicalFeatureType | It refers to the type of a physical feature. | cidoc-crm:E55_Type |
| PlaceType | It refers to the type of a place. | cidoc-crm:E55_Type |
| PlanType | It refers to the type of plan. | cidoc-crm:E55_Type |
| QualitativeValue | It refers to the qualitative characteristics of an observable entity, which are specified by special terms. | cidoc-crm:E55_Type |
| ConservationAndRestoration Intervention Decision-makingDecision | It refers to the final decision of the decision-making process. | cidoc-crm:E2_Temporal_Entity |
| ConservationAndRestoration Intervention Decision-makingIssue | It refers to the issue that the decision-making process aims to decide about. | cidoc-crm:E2_Temporal_Entity |
| ConservationAndRestoration Intervention Decision-making | It refers to the decision-making process which aims to decide for a conservation and restoration intervention. | cidoc-crm:E13_Attribute_Assignment |
| ConservationAnd RestorationIntervention | It refers to an implemented activity which aims to the preservation or/and the restoration of the object to the best possible condition, through the direct or indirect (i.e., environment) intervention to a conservation object (of any kind, movable or immovable CH). | cidoc-crm:E7_Activity |
References
[1]
H. Marcal, R. Macedo, A. Nogueira, and A. Duarte. 2013. Whose decision is it? Reflections about a decision making model based on qualitative methodologies. In Conservation, Exposition, Restauration d’Objets d’Art (CeROArt). Association CeROArt asbl. DOI:
[2]
C. Niang, C. Marinica, B. Bouchou-Markhoff, E. Leboucher, O. Malavergne, L. Bouiller, C. Darrieumerlou, and F. Laissus. 2017. Supporting semantic interoperability in conservation-restoration domain: The PARCOURS project. Journal on Computing and Cultural Heritage 10, 3 (2017), 16:1–16:20.
[3]
S. Muñoz-Viñas. 2012. Contemporary theory of conservation. Studies in Conservation 47, 1 (2012), 25–34.
[4]
J. Henderson and R. Waller. 2016. Effective preservation decision strategies. Studies in Conservation 61, 6 (2016), 308–323.
[5]
R. Letellier, W. Schmid, and F. LeBlanc. 2007. Recording, Documentation, and Information Management for the Conservation of Heritage Places: Guiding Principles. Getty Conservation Institute, Los Angeles, CA.
[6]
EN 16853:2017 Irish Standard. Conservation of Cultural Heritage - Conservation Process - Decision Making, Planning and Implementation. NSAI Standards.
[7]
M. Doerr. 2009. Ontologies for cultural heritage. In Handbook on Ontologies, International Handbooks on Information Systems. S. Staab, R. Studer (Eds.), Springer, Berlin, 463–486. DOI:
[8]
B. Appelbaum. 2007. Conservation Treatment Methodology. Routledge, London.
[9]
C. Niang, C. Marinica, E. Leboucher, L. Bouiller, and C. Capderou. 2015. An ontological model for conservation-restoration of cultural objects. In Proceedings of the Digital Heritage, Vol. 2, 157–160. DOI:
[10]
S. Michalski and M. Rossi-Doria. 2011. Using decision diagrams to explore, document, and teach treatment decisions, with an example of their application to a difficult painting consolidation treatment. In Proceedings of the ICOM-CC 16th Triennial Conference. J. Bridgland (Ed.), 1–8
[11]
T. J. Turk, A. M. Pranjić, A. Hursthouse, R. Turner, and J. J. Hughes. 2019. Decision support criteria and the development of a decision support tool for the selection of conservation materials for the built cultural heritage. Journal of Cultural Heritage 37 (2019), 44–53. DOI:
[12]
DOCAM. Retrieved March 17, 2024 from https://www.docam.ca/en/restoration-decisions/a-decision-making-model-the-decision-tree.html
[13]
Compatibility in Conservation Interventions. An IT Tool Developed by the PRODOMEA Project. Retrieved March 17, 2024 from https://www.academia.edu/34029617/Compatibility_in_conservation_interventions_An_IT_tool_developed_by_the_PRODOMEA_project
[14]
T. Hellmund, P. Hertweck, D. Hilbring, J. Mossgraber, G. Alexandrakis, P. Pouli, A. Siatou, and G. Padeletti. 2018. Introducing the HERACLES ontology semantics for cultural heritage management. Heritage 1, 2 (2018), 377–391. DOI:
[15]
F. M. La Russa and C. Santagati. 2021. An AI-based DSS for preventive conservation of museum collections in historic buildings. Journal of Archaeological Science: Reports 35 (2021), 102735.
[16]
E. Moraitou, Y., Christodoulou, and G. Caridakis. 2023. Semantic models and services for conservation and restoration of cultural heritage: A comprehensive survey. Semantic Web Journal 14, 2 (2023), 261–291.
[17]
AIC Wiki. BPG Surface Cleaning. Retrieved March 17, 2024 from https://www.conservation-wiki.com/wiki/BPG_Surface_Cleaning
[18]
AIC. BPG Consolidation, Fixing, Facing. Retrieved March 17, 2024 from https://www.conservation-wiki.com/wiki/BPG_Consolidation,_Fixing,_and_Facing
[19]
S. W. Michalski. 2018. Sharing conservation decisions: Tools, tactics, and ideas. In Sharing Conservation Decisions: Current Issues and Future Strategies. A. Heritage, J. Copithorne (Eds.), International Centre for the Study of the Preservation and Restoration of Cultural Property (ICCROM), Rome, 183–202.
[20]
A. Kupczak, M. Jędrychowski, M. Strojecki, L. Krzemień, Ł. Bratasz, M. Łukomski, and R. Kozłowski. 2018. HERIe: A web-based decision-supporting tool for assessing risk of physical damage using various failure criteria. Studies in Conservation 63, sup1 (2018), 151–155. DOI:
[21]
O. Schalm, A. Cabal, W. Anaf, D. L. Pernia, J. Callier, and N. Ortega. 2019. A decision support system for preventive conservation: From measurements towards decision making. The European Journal Plus 134 (2019), 74. DOI:
[22]
CIDOC CRM. Retrieved March 17, 2024 from http://www.cidoc-crm.org/
[23]
CIDOC CRM. Compatible Models & Collaborations. Retrieved March 17, 2024 from http://www.cidoc-crm.org/collaborations
[24]
M. Doerr, J. Hunter, and C. Lagoze 2003. Towards a core ontology for information integration. Journal of Digital Information 4, 1 (2003). Retrieved from https://jodi-ojs-tdl.tdl.org/jodi/article/view/jodi-109
[25]
Linked Conservation Data. Retrieved March 17, 2024 from https://www.ligatus.org.uk/lcd/
[26]
A. Zerbini. 2018. Developing a heritage database for the Middle East and North Africa Journal of Field Archaeology 43, 1 (2018), 9–18.
[27]
V. A. Carriero, A. Gangemi, M. L. Mancinelli, L. Marinucci, A. G. Nuzzolese, V. Presutti, and C. Veninata. 2019. ArCo: The Italian cultural heritage knowledge graph. In The Semantic Web–ISWC 2019: 18th International Semantic Web Conference, Lecture Notes in Computer Science, Vol. 11779, pp. 36–52.
[28]
S. Odat. 2011. A Semantic e-Science Platform for 20th Century Paint Conservation. Doctoral thesis. The University of Queensland, School of Information Technology and Electrical Engineering.
[29]
R. Cacciotti, J. Valach, P. Kuneš, M. Čerňanský, M. Blaško, and P. Křemen. 2013. Monument damage information system (MONDIS): An ontological approach to cultural heritage documentation. In Proceedings of the ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences II-5, 55–60.
[30]
M. Acierno, S. Cursi, D. Simeone, and D. Fiorani. 2017. Architectural heritage knowledge modelling: An ontology-based framework for conservation process. Journal of Cultural Heritage 24 (2017), 124–133. DOI:
[31]
T. Messaoudi, P. Véron, G. Halin, and L. De Luca. 2017. An ontological model for the reality-based 3D annotation of heritage building conservation state. Journal of Cultural Heritage 29 (2017), 100–112.
[32]
A. Zreik and Z. Kedad. 2021. Matching conservation-restoration trajectories: An ontology-based approach. In Research Challenges in Information Science (RCIS ’21). S. Cherfi, A. Perini, and S. Nurcan (Eds.), Lecture Notes in Business Information Processing, Springer, Cham, 415. DOI:
[33]
R. Wang and W. Chen. 2020. Case-based reasoning of damaged ancient buildings based on ontology. In Proceedings of the E3S Web Conferences, Vol. 165, 04007. DOI:
[34]
F. Boochs, A. Trémeau, O. Murphy, M. Gerke, J. L. Lerma, A. Karmacharya, and M. Karaszewski. 2014. Towards a knowledge model bridging technologies and applications in cultural heritage documentation. In Proceedings of the ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. II-5, 81–88.
[35]
Art & Architecture Thesaurus Online Full Record Display, Powdering. Retrieved March 17, 2024 from https://www.getty.edu/vow/AATFullDisplay?find=powdering&logic=AND¬e=&english=N&prev_page=1&subjectid=300379824
[36]
K. Zachila, K. Kotis, E. Paparidis, S. Ladikou, and D. Spiliotopoulos. 2021. Facilitating semantic interoperability of trustworthy IoT entities in cultural spaces: The smart museum ontology. IoT 2, 4 (2021), 741–760. DOI:
[37]
Google Sheets. Retrieved March 17, 2024 from https://www.google.com/sheets/about/
[38]
Google Docs. Retrieved March 17, 2024 from https://www.google.com/docs/about/
[39]
LOV. Retrieved March 17, 2024 from https://lov.linkeddata.es/dataset/lov/
[40]
ODP Main Page. Retrieved March 17, 2024 from http://ontologydesignpatterns.org/wiki/Main_Page
[41]
K. Kotis, G. Vouros, and D. Spiliotopoulos. 2020. Ontology engineering methodologies for the evolution of living and reused ontologies: Status, trends, findings and recommendations. The Knowledge Engineering Review 35 (2020), E4. DOI:
[42]
CRMact. Retrieved March 17, 2024 from https://cidoc-crm.org/crmact/node/8732
[43]
CRMinf. Retrieved March 17, 2024 from https://www.cidoc-crm.org/crminf/home-4
[44]
Decision Making Ontology. Retrieved March 17, 2024 from https://github.com/nicholascar/do-ont
[45]
CRMsci. Retrieved March 17, 2024 from https://cidoc-crm.org/crmsci/home-1
[46]
A. Velios, C. Meghini, M. Doerr, and S. Stead. 2023. Typed properties and negative typed properties: Dealing with type observations and negative statements in the CIDOC CRM. Semantic Web Journal 14, 2 (2023), 421–441.
[47]
CRMntp. Linked Conservation Dara—GitHub. Retrieved March 17, 2024 from https://github.com/linked-conservation-data/crmntp
[48]
SKOS Simple Knowledge Organization System—Home Page. Retrieved March 17, 2024 from https://www.w3.org/2004/02/skos/
[49]
Protégé. Retrieved March 17, 2024 from https://protege.stanford.edu/
[50]
D. Garijo. 2023. WIDOCO: A Wizard for Documenting Ontologies (v1.4.19). Zenodo. DOI:
[51]
SWRL: A Semantic Web Rule Language Combining OWL and RuleML. Retrieved March 17, 2024 from https://www.w3.org/Submission/SWRL/
[52]
SWRL Tab Plugin. Retrieved March 17, 2024 from https://github.com/protegeproject/swrltab-plugin
[53]
Pellet: An Open Source OWL DL Reasoner for Java. Retrieved March 17, 2024 from https://github.com/stardog-union/pellet
[54]
SPARQL Query Language for RDF. Retrieved March 17, 2024 from https://www.w3.org/TR/rdf-sparql-query/
[55]
Snap SPARQL. Retrieved March 17, 2024 from https://github.com/protegeproject/snap-sparql-query
[56]
Cellfie Plugin. Retrieved March 17, 2024 from https://github.com/protegeproject/cellfie-plugin
[57]
MappingMaster, Retrieved March 17, 2024 from https://github.com/protegeproject/mapping-master
Recommendations
Constructing a semantic wiki for living cultural heritage: case study of Korean tea culture
IMCOM '15: Proceedings of the 9th International Conference on Ubiquitous Information Management and CommunicationIn recent years, the amount of digital content related to cultural heritage resources has been growing to such a degree that it can now be easily published and distributed. However, such digital content is limited to material manifestations, such as ...
Analysis on Influence Factors of Public Participation in Cultural Heritage Conservation a Case of Kaiping Diaolou in Guangdong, China
ICEE '12: Proceedings of the 2012 3rd International Conference on E-Business and E-Government - Volume 04In China, cultural heritage conservation is mainly dominated by government officials and experts. The public are not only the creators of cultural heritage, but also the owners and inheritors of cultural heritage. Therefore, it is necessary for the ...
Cultural Heritage Representation in Databases
BCI '15: Proceedings of the 7th Balkan Conference on Informatics ConferenceAdvanced research means in the study of cultural heritage are more and more popular and used at large scale. Either there are X-ray installations, nuclear facilities or photonics equipment, these innovations proved to offer enough resources to study the ...
Comments
Please enable JavaScript to view thecomments powered by Disqus.Information & Contributors
Information
Published In
September 2024
382 pages
Copyright © 2024 Copyright held by the owner/author(s).
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike International 4.0 License.
Publisher
Association for Computing Machinery
New York, NY, United States
Publication History
Published: 22 May 2024
Online AM: 27 March 2024
Accepted: 29 February 2024
Revised: 08 January 2024
Received: 26 September 2023
Published in JOCCH Volume 17, Issue 3
Check for updates
Author Tags
Qualifiers
- Research-article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 808Total Downloads
- Downloads (Last 12 months)808
- Downloads (Last 6 weeks)209
Reflects downloads up to 25 Nov 2024
Other Metrics
Citations
View Options
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in