An ontology for the knowledge management of earthen constructions: the mix design of adobes

Chiara Cirinnà

University of Florence, Italy (email: chiara.cirinna@taed.unifi.it) Saverio Mecca

University of Florence, Florence, Italy (email: saverio.mecca@unifi.it) Marco Masera

University of Florence, Florence, Italy (email: marco.masera@unifi.it)

Abstract

The theory of Knowledge Management (KM), starting from the belief that knowledge is the most relevant source of competitive advantage, studies and provides methodologies and tools capable of managing it, through an innovative approach. It is recognized that knowledge dissemination and decentralization are the most important key points to improving the success of the proper use of a specific technique. The heritage of earthen architecture is in danger because it is particularly subjected to the destructive action of environmental agents, lack of maintenance, progressive abandonment and, above all, dispersion and losing of implicit knowledge. In this field, KM is a complex paradigm, because all the tacit knowledge consists of that kind of informal technical skills captured in the term “know-how”. Existent explicit knowledge is, instead, disorganized and dispersed. Moreover, knowledge has to be shared and disseminated, in order to be truly effective and profitable for the advancement of the research in the field. The goal of the work presented is the implementation of advanced methodologies for collection, management and sharing of knowledge on earthen building technologies through the application of ontology-based tools. The implemented tool is expected to be a valuable support for the analysis and the management of different sources of domain knowledge, and to enable its reuse and sharing among the operators involved in the construction process. Keywords: Knowledge management, ontology, architectural heritage.

1. Introduction

This paper is based on a Doctorate research focussing on the application of ontology-based tools to the management of knowledge of earthen construction techniques. Earthen architecture is characterized by a high level of technical variability and integration into its geographical and technical environments. The tradition of building with earth is spread the world over and in Mediterranean regions it covers a large geographical area, with a variety of typologies and vernacular expressions that reflect the complex cultural identity of the Mediterranean basin. During the last decades

Earthen Architecture technologies have been acknowledged as a research field comparable with the other building technologies and interest has increased not only in conservation practices but also in new earthen construction [1]: at present, earthen building technologies have several kind of applications. The heritage of earthen architecture is in danger for two kinds of reasons: first, it is particularly exposed to deterioration, lack of maintenance and progressive abandonment; second, it is subjected to the dispersion and loss of implicit knowledge. The second issue is certainly the most dangerous. Recently the growth of awareness regarding the risk of losing this kind of knowledge has contributed to the awakening of a consciousness and to the beginning of a debate.

2. Environment of the research

Clay is the world's oldest mineral building material. The reasons that justify the use of earthen building technologies in the Third Millennium are related to several factors. First, the high performances: the use of earth allows to reduce noticeably the energetic consumption for air-conditioning and to reach considerable acoustic insulation, thanks to the inertia due to the high mass of building structures. Moreover the use of earth allows the purification of the internal air and the control of humidity. Second, the environmental sustainability: the whole production cycle followed for new constructions entails a very low consumption of raw materials and energy, and low emissions of pollutant and discard. Third, the versatility of applications: earth can be used with different building techniques, in relation to the geographic specificity, to the availability of tools and raw materials and can be used by small and large construction companies. Notwithstanding, there are some elements that hurdle the diffusion of this building technique. Some of the main issues can be summarized as follows:

• the information specific to the field of conservation of earthen architecture is lacking in a number of areas. Nevertheless, a great deal of information exists in grey literature that is not readily accessible or reported on. Additional information exists in other disciplines, such as clay mineralogy, geotechnics, soil mechanics, microbiology, pedology, seismic engineering, thermal engineering, etc., but has not been informatively adapted and applied to earthen architecture;

• for each specific research subject, it will be necessary to do a preliminary cross-disciplinary desk assessment and synthesis. Identified missing pieces will have to be adopted/adapted from other disciplines or generated specifically as a result of a specific research effort. This involves not only engaging researchers and professionals from other disciplines, but also allowing researchers within the earthen architecture conservation field to devote time to exploring these ancillary areas of

study, in order to foster the necessary links and transfers;

• the access to existing data in the field of earthen architecture and its conservation is severely limited due to the lack of centralized databases, online catalogues meeting international library standards, and the like. Increased access is key to fostering research that is responsive to needs in the field;

• the results of the research should be made broadly accessible, especially through the development of appropriate diagnostic methods and analytical tools.

The scenario described above leads to some considerations: the lack of applications is strictly connected to the fundamental issue of managing and organizing the existent scientific knowledge and the local and tacit technical knowledge, and to integrate it with national and international references from other disciplines, such as standards or recommendations. Moreover, in this field, it the opportunity of enabling sharing, reuse and updating of the knowledge base is critical: the knowledge obtained must be well maintained and up-to-date. This is essential for ensuring that decisions are based on the latest information available. The design of new earthen constructions requires a deep knowledge of physical and mechanical characteristics of the building material [2], as well as of building procedures, on which depend quality and features of end products. The phase of selection of the appropriate earth is decisive in order to ensure the production of high-quality buildings. For this reason the research has been focussed on the analysis of mix design of adobes, that are unburned bricks made of clay which can be mixed with cellulose fibre or other materials to create a durable building material. The purpose of the Doctorate research aims at experimenting a new KM tool that enables one to manage the technical knowledge and to develop a methodology for supporting decision processes of the actors involved in the construction process: the attention is posed on the application of the ontology methodologies derived from the field of the artificial intelligence.

3. What are ontologies

In recent years the development of ontology is playing a key role in knowledge management. The term ‘Ontology’ has its origins in metaphysics and philosophy. In this context, ontology is used to explain the nature of the reality. In computer science, there are at least a dozen definitions of ontologies in the literature [3]. One of the most recent ones states that an ontology provides the skeletal knowledge and an infrastructure for integrating knowledge bases at the knowledge level, independent of particular implementations [4]. This definition assumes a particular importance in this research in terms of the necessary mechanisms to represent, share, and reuse the existing knowledge of an earthen architectural heritage domain.

Why is there now renewed interest in ontologies? “Ontology” is the name given to a structure of knowledge, used as a means of knowledge sharing within a community. An ontology is a “data skeleton” that can be developed for a variety of data-based and knowledge-based systems. It consists of a collection of concepts (or classes) and relationships between them. Generally, each concept is represented by a favoured term, word or phrase. The ontology might include principle synonyms (abbreviations, common misspellings) for each term. It could contain only a single type of relationship between terms; e.g., a “kind-of” relationship, such as “pisé and adobes are kinds of earthen building techniques.” However, ontologies can represent other relationships (e.g., part-of, precedes-in-process, reports-to). Further, terms in ontologies can have qualifying attributes, and ontologies can include restraints (“rules”) on values for terms. So, an ontology includes basic knowledge relationships that a variety of types of knowledge applications might use. For humans, ontologies enable better access to information and promote shared understanding. For computers, ontologies facilitate comprehension of information and more extensive processing. In some sense, computers have been using rudimentary ontologies in the form of data dictionaries, enterprise data schemas, web architectures and taxonomies. As systems begin making fuller use of ontologies, computers can make sense of unstructured and semi-structured materials and take on a significantly more extensive role in processing transactions because they ‘know’ how a piece of information (document, fact, record) relates to other pieces of information.

Complexity level

Catalog

Glossary Taxonomy

ThesaurusA collection of

frames

General Logicalconstraints

Figure 1. Different levels of ontological complexity

In summary, ontologies are about adding meaning to information in order to provide clear value, delivering “the right information to the right person at the right time and in the right way.” The main goals associated to the development of the ontology-based tool are [5]:

• to analyse the domain of the existent fragmented body of knowledge: relevant pieces of knowledge are assumed to contribute to the effectiveness of proper building procedures;

• to share a common consensual knowledge among all the actors involved, i.e. the

researcher, the architect, the conservation scientist, the conservator-restorer and the end users;

• to leverage tacit knowledge related to the identified building procedures, in order to store dispersed and unstructured domain knowledge, such as domain competencies characteristics, project experiences and contextual knowledge;

• to separate domain knowledge from the operational knowledge, like the ability to perform context-based information retrieval, presenting contextual or situational knowledge about information sources and assisting the execution of appropriate preventive actions and problem solving tasks

• to enable reuse of domain knowledge: the creation of a common vocabulary facilitates communication in design and maintenance issues across people with different professional backgrounds. For example, the ontologies can be reused in several domains, since the concepts they define are likely to be used almost universally.

For the research, the possibility of using a knowledge model of Protégé 2000, an ontology editing and knowledge-acquisition environment has been assessed. Protégé-2000 is the latest component-based and platform-independent generation of the Protégé toolset. It has been developed as a small application designed for a medical domain. At present, it has become a general-purpose set of tools for building knowledge-based systems.

4. The methodology

At present, even if ontologies have been tested and applied on several domains, a unified methodology for the development of ontologies is missing and scientific literature does not provide publications on this issue. The development of an ontology is a hard and iterative process and often time-consuming. Therefore I find appropriate to describe the basic steps followed during the research. The methodology adopted for the ontology development comprises the following phases:

1. Identification of the research topic. The preliminary step is characterised by the identification of the domain through a complete and accurate definition of the “boundaries” of the topic selected. This evaluation is considered crucial for the effectiveness of the ontology: the more the domain is defined, the more the ontology will be clear and effective and will constitute a valuable support for the organisation and the retrieval of relevant knowledge. The identification of the domain is based on the definition of the purposes to which the ontology will be used. As suggested by Gruninger e Fox [6], it could be useful to outline a list of questions that the ontology must answer (competency questions). For the mix design domain these are possible

competency questions: Which are the most relevant properties of the earth in order to evaluate their suitability? What kind of analysis do I have to carry out for obtaining a right evaluation of their properties? Has the different kind of clays a direct influence on the performances of the earth? These questions are critical also during the final step of the ontology development, when it will be necessary to evaluate the completeness of the information recorded inside the knowledge base. The ontology of mix design aims at organising a double-sided field of knowledge representation. On the one hand it aims at reorganising the fragmented and dispersed domain of knowledge pertaining to the mix design of adobes, using therefore “descriptive” information that are able to describe the topic from different point of view (elements that constitute the material, tools, actors, etc); on the other hand the ontology intends to support the decisional actions needed for the definition of the appropriate mix of earth for the production of adobes, involving “operational” knowledge, as such necessary for the development of procedures and tasks.

2. Review of existing ontologies/taxonomies. The use of ontologies can provide a significant advancement of knowledge among different communities. For this reason, it is important to evaluate the opportunity of reviewing existing ontologies or taxonomies, in order to reuse the related knowledge, through the improvement and enlargement of the information contained in the ontology. The reuse of existing ontologies/taxonomies can provide an added value if the system implemented needs to interact with other applications supported by a particular ontology or a specific vocabulary. Many ontologies are developed in an electronic format and could be easily imported and integrated [7]. Before the development of the ontology of mix design, a research on existing ontologies has been performed. Unfortunately no relevant ontology or taxonomy has been found.

3. Identification of sources of domain knowledge. One of the main problems that characterised the initial step of the ontology development was related to the heterogeneity and dispersion of information. A huge work has been accomplished in order to reorganise the knowledge domain that was not ready to be formalised. Therefore it was necessary an initial phase of harvesting, elaboration and selection of information, essential for the reorganisation of mix design related knowledge. This step entailed the gathering of different typologies of knowledge, information and data. It is useful to provide a definition of the above three terms, since they are have similar meaning but different content [8]. Data is the raw material of information; a single piece of data has no meaning unless the context is understood. Data needs to be transformed to information. Information is data that has been given meaning by way of relational connection to a context or a specification. Knowledge is composed of experience, values, insights and contextual information. It is the appropriate collection of information, such that it's intent is to be useful and to allow the execution of a decision or an action. According to the vision of Nonaka and Takeuchi [9] these kinds of

knowledge are characterised by different degree of formalisation, since they can be tacit or explicit. It was interesting to discover that the lower was the level of formalisation (tacit knowledge) the richer was the kind of knowledge available. The information needed to develop the ontology was acquired from different sources with various methodologies for extraction and management. They can be summarised as follows:

a. Explicit information and knowledge. - Technical explicit knowledge deriving from treaties and manuals, containing information about the specific field of earthen constructions: the knowledge contained is highly formalised, so it has been used without operating further elaborations;

- Technical explicit knowledge deriving from proceedings of national and international symposiums and scientific reports dealing with the specific sector of earthen constructions: in this case knowledge has been elaborated, formalised and adapted to the objectives of the ontology;

- Technical explicit information obtained from normative and procedural data, pertaining to other application sectors that developed earth analysis procedures, but for different purposes (e.g. road foundation construction or soil analysis);

- Technical explicit information pertaining to various sectors (geotechnics, chemistry, mineralogy etc.), necessary for organising the domain, obtained from other domains literature: publications, manuals, technical reports.

b. Tacit knowledge. - Tacit knowledge obtained through interviews to researchers and professionals coming from different fields. Thanks to the interaction with these experts, the relevant concepts necessary for the description of the domain have been characterised and a great part of tacit knowledge has been acquired.

- Tacit knowledge obtained from semi-structured interviews directed to local builders and craftsmen. The greatest difficulty was to obtain a story from the people interviewed, that are used to execute their action rather than talking about them. The acquisition of knowledge has been accomplished through the observation of their skilled gestures.

c. Data. - Data have been gained from laboratory tests, executed on the soils selected from craftsmen that have been asked to locate some types of soils considered appropriate for constructions. The analysis of these data have been interpreted in order to acquire knowledge that did not emerge from interviews.

4. Elaboration of information. The different types of information have been

selected and processed, in order to keep only those really necessary for the construction of the domain. Afterwards this information has been represented inside the ontology through the construction of a taxonomic structure that is containing the concepts used for the domain modelling. It is important to underline that the objective of the ontology is not the organisation of a complete vocabulary with all the concepts pertaining to the domain, but rather the development of a domain taxonomy containing the most relevant and useful concepts. Hence the objective of this step was to capture the main entities that are representative of the domain with the related relationships and properties. This phase is necessarily an iterative process, subjected to several modifications and improvements: the elaboration starts from an initial taxonomy that contains concepts derived from a preliminary brainstorming, necessary for obtaining all the possible classes of entities.

5. The ontological representation of concepts. Once defined the top level classes, it is required to explicit the individual classes through the taxonomic representation of concepts, based on subsumption rules. Every single concept can have subclasses, more specific than the superclass, following the relation “kind-of”. After having completed the taxonomy, the next step is about the definition of the relationships among classes and the properties that characterise each class. Indeed the classes, organised following a taxonomic structure, are not sufficient to model a domain. The added value of ontologies is represented by the possibility of specifying the relationships that exist among classes, by defining for each relationship the attributes or properties that characterise it. Every relationship has a definite meaning, it establishes the admissible values and the concepts to which refer to, it specifies the imposed constraints. Properties can have different features: they can describe a relationship between two classes and so connecting for instance a class “Procedures” to a class “Documents” by the relationship ”described_by” or they can represent a definite an alphanumeric or Boolean (true/false) or qualitative (red, blue or green) value, etc.

6. Elaboration of the knowledge base. After having defined the architecture that governs the ontology, it is required to populate it with examples or, to be more precise, with concrete instances of the information. The instances constitute a true knowledge base that contains the description of the procedures, the tools used, the materials, the actors involved and it is obtained through the elaboration of data, information, tacit and explicit knowledge in the form of descriptions, images, tables, graphs, classifications, etc.

4. The mix design ontology

The system arranges the single data classes in a taxonomic structure, that can be

described starting from the top level classes.

Figure 2. The top level taxonomy

The superclass “Thing” is the most general class that the system arranges automatically to start building the ontology. The top level classes and the main semantic areas that compose the ontology are “Materials”, “Tools”, “Actors”, “Conceptual elements”, “Protocol elements”.

“Materials”: this class includes the description of natural or manufactured materials, used or produced by the procedures. It is composed of all the natural and artificial materials that are used during the procedures for characterisation of mix design.

“Tools”: this class consists of the physical objects created by men, necessary for the development of the procedures. They are grouped in subclasses that refer to the single procedure considered.

Figure 3. The upper classes “Materials” and “Tools”.

The class “Actors” includes people, both individuals and groups, that perform intentional actions on their own responsibility in the sector of mix design selection procedures. They include organisations or associations that operate collectively and

individuals that work in the field.

Figure 4. The classes “Actors”, “Conceptual elements” and “Protocol elements”.

The class “Conceptual elements” contains all the immaterial entities that are produced by intellectual activities. It is composed of two main subclasses: the class “Procedures” contains the methodologies for the development of activities related to the selected domain. They are described following the scientific literature (codified procedures) or the practical skills of vernacular expressions (non codified procedures). The class “Document” collects all the descriptive elements that help in representing the domain: they are “codified documents” if related to knowledge deriving from bibliographies, thesauri, laws, instead they are “non codified documents” if they are produced by individuals and so they are not official documents.

These classes constitute the descriptive part of the ontology, since they organise the knowledge necessary to define the general framework, composed of data, information and knowledge that must describe the domain. This part of the ontology can be investigated independently from the rest of the system as a detailed and organised archive of information. The fifth upper class, “Protocol elements”, unlike the previous ones, contains all the knowledge necessary for the development of the activities, in order to obtain specific qualitative results, in a way that can be used as an operative tool for supporting decisions. It is composed of three direct subclasses: “Elementary activities” that describe the single activities executed by actors that produce inputs for the development of subsequent actions, “Procedures” that organise in a time sequence all the operations that were organised in a taxonomic way in the previous class, and “Process” that reconstructs the whole process to be followed during the mix design

selection procedures. The class “Process” is the focal point of the ontology, since it is the supporting tool for the development of the procedures. This class models the entire process, where the hexagons are the single procedures that can be explored and the arrows represents the various relations that connects the procedures that can be of various types like causal, temporal, spatial. Each procedure houses different tools that help the development of the process (spreadsheets, charts, patterns).

Figure 5. The class “Process” that organises the whole mix design procedures.

The resulting knowledge base then can be accessed using a problem-solving method to answer questions and solve problems regarding the domain. This ontology can be used for supporting decision when approaching new building projects, suggesting the best practice in a particular area with a definite kind of earth and, trough the system of queries, answering questions of the researcher, the architect, the conservation scientist and the end users. Furthermore, one of the main motivation behind ontologies is that they allow for sharing and reuse of knowledge bodies in computational form: a flexible access to a “shared workspace”, could be represented by a shared server on internet which supports document upload, event notification, group management and much more. To access a workspace, group-members only need a standard Web browser and a login-name and password.

5. Conclusions

The debate that has developed in the last few years on cultural heritage conservation has underlined the priorities to be faced, in order to stop the progressive disappearance of the earthen building techniques. The most important international organisations on the sector have expressed the urgent need to define new tools and operational strategies that enable actors to operate at the local level with more effective and specific conservation actions. The Knowledge Management technologies have been considered as the main references to identify innovative strategies for the management of processes for the conservation of monumental sites.

The Doctorate research presented in this paper evaluated the potentiality of ontologies as operational tools for creating, storing and managing valuable knowledge and make it usable by the actors involved in earthen building processes. This work has to be considered as an endeavour in this direction and as a base for future research.

References

[1] Balderrama A.A. (2001) The Conservation of Earthen Architecture, The GCI Newsletter, Vol.16

[2] Brown P. W., Clifton J. R., Robbins C. R. (1978) Methods for characterizing adobe building materials, In: NBS (National Bureau of Standards), Technical note, vol. 977, U.S. Government Printing Office, Washington. [3] Uschold M., Gruninger M. (1996) Ontologies: Principles, Methods and Application. Knowledge Engineering Review, Vol. 11, No. 2, pp.93-115. [4] Fensel D. (2000) Ontologies: Silver Bullet for Knowledge Management and Electronic Commerce, Springer-Verlag. Berlin. [5] Noy N. F., McGuinness D. L. (2001) Ontology Development 101: A Guide to Creating Your First Ontology, Stanford University. Stanford.

[6] Gruninger M., Fox M.S. (1995) Methodology for the Design and Evaluation of Ontologies. In: Proceedings of the Workshop on Basic Ontological Issues in Knowledge Sharing, IJCAI-95, Montreal. [7] El-Diraby T. A., Lima C., Feis, B. (2005) Domain Taxonomy for Construction Concepts: Toward a Formal Ontology for Construction Knowledge, Journal of Computing in Civil Engineering, Vol. 19, No. 4, pp. 394-406. [8] Brooking A. (1999) Corporate Memory, strategies for knowledge management, Thomson Business Press. [9] Nonaka I., Takeuchi H. (1997) The knowledge creating company, Guerini, Milano.