ICCD2OWL: an ICCD to OWL data compiler 1 A. Aiello, 1 S. Brandi, 1 M. Mango Furnari and 2 F. Proto 1 Istituto di Cibernetica “E. Caianiello”- CNR E-mail: {a.aiello, mf }@cib.na.cnr.it s.brandi@remuna.org 2 Dipartimento di Discipline Storiche “E. Lepore” Università Federico II di Napoli E-mail: fiona.proto@virgilio.it assets and those that describe the immaterial assets, and not least those related to the authority files and to the multimedia Abstract— This paper outlines the problems the authors documentation, have been investigated. Moreover, considering came across when, after the realization of ReMuNaICCD, an their importance in achieving a comprehensive description of ontology supposed to allow a more articulate use of the present the excavations, recognitions and archaeological essays, the cultural heritage information system, they approached the task cards that were structured by the ICCD in order to document of populating it with instances. To permit immediate use of the ontology, it was necessary to the stratigraphic units have also been studied and transferred develop ICCD2OWL, a data compiler able to translate the data into the ontology. recorded in the available ICCD cards into class instances and It was necessary to make a careful study of the ICCD cards. property instances for the ontology. The first task was to design First of all, in order to distinguish two macro-groups of formal codifications for both the ICCD attribute-value data fields: those giving information about the object the card is model and the ReMuNaICCD ontological data model. Then a program for converting each of them into the other was devoted to, and those giving information on the card itself. required. Inside these two macro-groups, we have characterized the Actually, the issue faced in this paper comes under the wider fields designed to supply identifications and characterizations, topic of the mappings between simple attribute-value data the fields designed to report events, and the fields that structures and ontological representations of the knowledge or, determine relations with other elements (documents, other in more general terms, between not-object oriented models and assets, other cards etc). object oriented models. The most remarkable conclusion is that part of the effort In order to establish which classes had to be created, we required is due to the semantic inadequacies of the ontology have distinguished on one hand, fields that appear without languages currently available. Such inadequacies make any variations in all ICCD cards, from those designed for taking attempt at accomplishing semantic interoperability based on into account of specific typologies of assets. On the other those languages alone unproductive. Therefore, it is evident hand, we have determined, for every field, whether it was that the methodologies and results presented here could profitably be used in a more general framework like the directly or indirectly related to the subject catalogued by the Semantic Web. card. ReMuNaICCD is not in any sense a simple ICCD fields I. INTRODUCTION transposition into an object oriented model. In order to In designing ReMuNaICCD, a formal ontology for the represent all the aspects of the cultural assets and the events Cultural Heritage domain [1], we had as fundamental reference that happened to them, it has proved to be a new data model, the recommendations issued by the Central Institute for the built from scratch. Catalogue and Documentation (ICCD). The research was To summarise, starting from the ICCD cards, which supported by the Virtual Museum of Naples project ReMuNa1 include 27 fields of the Bibliography card, to the nearly 300 (which stands for Network of Neapolitan Museums) and fields of the Architectural card, with an average of 200 fields SIABeC2 (Information System Applied to the Cultural per card, we elaborated ReMuNaICCD.v2.0, an ontology of Heritage). 381 classes (199 are Appellation subclasses), 473 This project was motivated by the necessity of having an objectProperties, 458 dataTypeProperties and about 750 ontology model compatible with the information system instances (nearly all of them were taken from the ICCD currently deployed inside the institutions that are in charge of vocabulary). the safekeeping, maintenance and valorisation of the cultural ReMuNaICCD has been written using the subset of OWL called OWL Lite [2]. OWL is the acronym of Web Ontology assets. Language [3], a standardized language for the specification of To this end, the ICCD cards that describe the material formal ontologies, recommended by the W3C. The main factor in choosing OWL Lite was the ascertainment that not 1 The ReMuNa project is financed by the MIUR with the Law n.488 only it offers a sufficient expressivity, but also guarantees a initiative of Cluster priori computational tractability of the final product [4]. 2 The works [5] [6] and [7] of Guarino et al. on the formal The project SIABeC is financed with the projectCentro regionale di ontologies foundations, those of Gangemi et al. on the Competenza per lo sviluppo ed il trasferimento della innovazione tecnologica (INNOVA) P.O.R. Campania misura 3.16. ontology patterns [8], the guide lines proposed by Alan Rector et al. [9] and the newest experiences reported online by the subclasses: OEP [10], have strongly influenced the development of our The E n d u r a n t , i.e. the class of the subjects/objects ontology. represented by abstracting them from all possible contingent Since the start of our activities in this field, we have used considerations, namely, attributes and relations that are Protégé as the only tool for editing our ontologies. Protégé is understandable if and only if they make reference to certain a free, open source ontology editor and knowledge base specific space-time contexts. framework [11] [12]. Furthermore, we directly linked Protégé The Perdurant, the class of the observations in which the and the system Sesame [13] [14] for the exchange of subjects/objects emerge in a given space-time context, and ontologies through the Internet, by means of a plug-in [15] exhibit a set of attributes and relations that are specific to that produced inside the project ReMuNa [16]. context. In section II, we introduce the main features of the The Space-Time Region, the class that models the space ontological model: the uppermost components of the class and time where the observations are detected. hierarchy and the fundamental ontology pattern. Section III Besides the subsumption relation, that determines the gives the basis for the development of the compiler. In the classes’ hierarchy, in an ontology, classes are interlinked first subsection, the input and the output data models are through a network of specific relations that formalizes not compared; in the second subsection, it is explained how to use only the individual meaning of each class but also, and more the Historicism Pattern in order to get “Conceptual interestingly, the meaning of their staying together. Translations”. Section IV, in five subsections, describes the It has been discovered by Gangemi et al. [8], that some not formalisms adopted in order to a) serialize the ontology trivial “conceptual structures” of the domain prove to be patterns, b) identify the instances, c) codify the classes, d) use represented by “semantically autonomous” clusters inside the global variables, and e) define a configuration file. Section V ontological network, each involving many classes and many outlines the processes that the compiler performs when properties, and sometimes even meta-properties (properties of translating ICCD to ReMuNaICCD and viceversa. In Section properties). Those “conceptual structures” are totally lost when VI, some conclusions are summarized and a question is asked: the ontology is coded using the currently available languages aren’t the current ontologies' formalizations absolutely for ontology, namely RDF/RDFS and OWL. Indeed, those inadequate for solving the problem of semantic languages do not offer any syntactical constructs that permit interoperability? the representation of ontological substructures, and even less the essential accompanying instructions for their use. II. THE ONTOLOGICAL MODEL Currently, ontologies are represented as oriented graphs in The ontology ReMuNaICCD has been built taking account which the nodes represent the classes and the arcs represent the of the most common top-level ontologies, in particular properties. Therefore, any “conceptual structure” relevant to DOLCE [17] and ICOM/CIDOC-CRM [18]. Here, we the domain will be represented by some pattern of connected introduce only a small part of the ReMuNaICCD classes and node-arc-node triples abstracted from the whole graph. properties, just those that are involved in the topic dealt with Actually, ontology patterns, made of classes and correlating in this paper. properties, constitute an optimal instrument for highlighting The root of the classes’ hierarchy is the class Entity. It is common features and for generalizing complex relationships or the class that represents the universe we are interested in and is decomposing them in their constitutive elements. articulated in two radically different subclasses (Fig. 1): The Historicism Pattern illustrated in Fig. 2 gives the most Concrete: the more general class that comprises all the important conceptual structure in the ontology ReMuNaICCD. concepts that model the domain that we have to analyze and to It shows how the various kinds of historical reports are formally represent. represented in terms of the fundamental classes, i.e. Endurant, Perdurant, and Space-Time Region and some Entity sub properties of comprises, the most general objectProperty that formalizes the specific tools for performing the analysis of subclass_of Concrete. Concrete Appellation In this ontology pattern, Participation, a subclass of Perdurant, plays the role of the basic atomic module by subclass_of means of which all the classes representing the various kinds Endurant Perdurant Time-Space Region of historical reports can be built. The constructors are some (non-transitive) sub properties of comprises. Participation formalizes the observation of a single object (has_present = Fig. 1. The ReMuNaICCD’s top-level ontology Endurant) while it plays a certain role (has_role = Role), in participating in an elementary interaction (participates_in = Appellation: the root of the dictionaries that were created H i s t o r y _ F r a g m e n t ), at a certain place and time and are controlled by third parties, whose semantics remains (has_space-time_location = Space-Time_Region). alien to ReMuNaICCD.Concrete is divided into three disjoint sequence of attribute-value pairs concerning the Cultural Heritage. The admissible values of the attributes depend on Concrete the cultural asset in question. In any case, the assignment of comprises values to attribute is not always compulsory, just as repeated subclass_of assignments of single independent values or groups of values has_space-time_ can be permitted. Endurant Perdurant location Space-Time Region Actually, the technical specifications of the ICCD, i.e. the sequential file format, and its intrinsic characteristics, i.e. the subclass_of arbitrariness of the number of the fields and the presence of has_ functional dependences among the fields, mean that the results present already reported in the literature, like many concerning Participation Historical_Period role relational databases, are irrelevant, making it necessary to start type again from scratch. participates characterizes Historical_Event The ontological structure described in the previous _in paragraph constitutes the basic skeleton of ReMuNaICCD. report_of That structure was enriched with classes, properties and History_Fragment instances deduced by the various segments (paragraphs, fields type and subfields) of which the ICCD cards are made up. It must be remarked that, the modalities for determining the Fig. 2. The Historicism Pattern relations in the ICCD schema are radically different from those used in the ontological model. An example of this can be Participation has a particular role because, through the given by comparing the modalities that the two models adopt primitive relation has_present, it embeds the Endurant into in order to deal with the relationship between the the space-time context (“Endurant present_in Perdurant”). subjects/objects and the events to which they participate. Indeed, Participation is the class that binds the single Let us consider a subject “X” that carries out one of the Endurant to the attribute and the relation (i.e. the properties following two functions: h a s _ r o l e and p a r t i c i p a t e s _ i n ) that describe its (a), the function of a scientific director in a survey which manifestation in the specific space-time context indicated by enabled the finding of a good “α” has_space-time_location. (b), the function of a collaborator responsible for compiling The class History_Fragment, a subclass of Perdurant, an ICCD card of the good “α” models the interactions. Namely, two or more instances of then, according to the ICCD data model: P a r t i c i p a t i o n are connected b y t h e property in the case (a), the name of the subject “X” has to be p a r t i c i p a t e s _ i n to the s a m e instance of reported in the simple fields of the ICCD card of the good History_Fragment: the Endurants that are present_in “α”, which refer to the Scientific Director of the Survey, those Participations are intended to interact with each other which have enabled the finding of the good “α”; playing the roles indicated inside their respective in the case (b), the name of “X” has to be reported in the Participations. fields that refer to the Collaborator responsible for the The class Historical_Event models observations which compilation of the ICCD card. are more complex than the single History_Fragment; many In ReMuNaICCD, instead, History_Fragment s can be composed in a single in the case (a), the subject “X” is an instance of Person, Historical_Event through the property is_report_of and present_in, an instance of Participation, with the its inverse has_report. attribute role set to Scientific_Director; that instance of Finally, every entity comprising whatever number of Participation p a r t i c i p a t e s _ i n an instance of Historical_Events is defined as Historical_Period. History_Fragment that is report_of an instance of The Historicism Pattern, through the property role of the Survey (subclass of Historical_Event); class Participation and the property type, defined for any in the case b), the subject “X” is an instance of Person Concrete, can describe the conceptual structure of any kind that, present_in a Participation with the role of of historical report, even reports which are much more C o l l a b o r a t o r , p a r t i c i p a t e s _ i n an instance of complex than those codified by the ICCD schema. History_Fragment of the type Compilation_of_ICCD_ card III. FROM ICCD TO REMUNAICCD A. Comparing the two data models B. The Conceptual Translations As already mentioned, the ICCD format3 consists of a After the foregoing analysis, which showed the basic conceptual elements of the ICCD recommendations, a 3 The technical specifications c a n be found at complex re-composition process becomes necessary. Firstly, http://www.iccd.beniculturali.it/standard/ we must find the correspondences between every ICCD field RSTN and RSTR, the data compiler has to build the instances and the elements that constitute our ontology. Then, every represented by the red boxes and, at the same time, implement segment of the ICCD cards must be associated with some the mapping given in Table 1: the contents of the fields in the ontology pattern that connects “semantically” its ontological first column become the values indicated in the second representative to the class describing the card’s main subject. column (references are made using the same symbols as Fig. Therefore, we need to design some formal languages (XML 3); the third column gives the classes of the instances related documents) that enable us to express: a) the correspondences to the values referenced in the second column. The other boxes ‘ICCD field – ontology element’, b) the ontology patterns, in the pattern represent either built-in instances (in yellow) or and c) how the compiler must use the previous data so that the instances (in black) that transpire from the translation of other mechanical translation of an ICCD card into classes, properties ICCD fields. and instances, does in fact lead to the expected codifications. To give an example of this process, let us consider the card for the archaeological find (RA) and its structured field ICCD Instance name Instance Class reserved to the restoration data. An archaeological find, field or property value or property reported in the card RA is dealt with by ReMuNaICCD as an RSTD TR Time_Region instance of the class Archaeological_Find (subclass of RSTS situation(of v2) Situation (of Restoration) Endurant), while a report about a restoration is dealt with as RSTE v4 Juridical Person an instance of R e s t o r a t i o n , a subclass of RSTN v5 Physical Person History_Fragment. The ontology pattern instance shown in Fig.3 describes how Archaeological_Find Table 1. Mapping ICCD fields vs ontological elements instances are related to the ontological correspondents of the ICCD fields devoted to restorations: RSTD (date), RSTS IV. AN OUTLINE OF THE ICCD2OWL FORMALISM. (situation), RSTE (responsible agency), RSTN (operator) and RSTR (financing agency). A. The pattern’s serialization To make a codification which is machine-understandable, we serialized all the patterns and all the rules concerning the has_present Arch_Find matchings into XML documents. Time-Region AFX In order to optimize the reuse, patterns and fields are defined TR Participation separately. role Role P1 v1 has_time_ location Binding has_present ICCD field/ RSTE Jur_Pers Pattern Arch_Find-Jur_Pers-name Restoration v4 History_Fragment Participation V2, SITUATIONY role Role Binding v1 P2 v3 Parameter/ ”Restored_object” has_participant value _in has_present Jur_Pers v2 v4 ”Restoration” (2) Participation Y role Role SituationY P3 v3 v3 ”Financial_Support” has_present Phy_Pers v4 v4 Participation @@RSTE@@ role Role P4 Start Class Archaeological_Find v5 Condition: TSK=”RA” TSK Fig. 3. Example of Historicism Pattern instance ”RA” Of course, in the graph, the boxes represent instances: the upper part of each box contains the name of the instance’s class and the lower part contains the name of the instance (in Table 2. Example of ICCD field/Pattern binding the case of Restoration, the value SituationY of the property situation is also reported). The patterns are referable by unique assigned identifiers so that each ICCD field can be bound to one or more precise To transfer the contents of the fields RSTD, RSTS, RSTE, patterns. Moreover, since the same ICCD field can be bound The foregoing policy is implemented by coding, in the to different patterns, depending on the context, our formalism configuration document of ICCD2OWL, the properties that also considers even the possibility of declaring parameters as mostly carry identity criteria and the rules according to which well as conditions on the choice of the bindings. starting from their values, the instances’ identifiers must be For example, inside the ICCD, there is the field Card composed. Type (TSK) that specifies the subject of the card, namely the C. Class use case specifications “starting class”. It can be an Archaeological Find or an Archaeological Site or whatever other cultural good or asset. It As we have already seen, the same class can be instantiated is clear, that on the basis of this piece of information, the for representing entities that receive the values of the same choice of the bindings proves to be restricted and the property from different ICCD fields. For example, an instance parameters that must be passed are determined. The code that of the class Physical_Person receives the value of name formalizes this choice, for the previous example, is sketched from the field RCGA (scientific advisory) if the ICCD card in Table 2. reports that entity as a Scientific Director. Differently, in The patterns are serialized as lists of triples whose the case the entity is reported in the ICCD card as an elements contain parameters that assume values according to Authority, then the instance of the class Physical_Person the fields of the ICCD cards. The lists may contain names of must receive the value of name from the field FUR (the other patterns that can be considered as sub patterns of those collaborator responsible). codified by the lists. For example, the codification of the pattern Arch_Find-Jur_Pers-name quoted in Table 2 is sketched in Table 3. Class name inside Physical_Person the config file Pattern to the instance’s ID @@PREFIX@@Phy_Pers_@@v1@@ Pattern Id Arch_Find-Jur_Pers (referred by v1) In param. v1,v2,v3,v4 Class name Arch_Find-@@v2@@ inside ReMuNaICCD @@PREFIX@@Physical_Person subpattern v1,v2 Out param. Juridical_Person-@@v2@@ Table 4. Example of the use case specification of a class v3,v4 For the correct interpretation of the ICCD cards, ICCD2OWL receives the information it needs in cases like the example above by means of pieces of code like the one shown Juridical_Person in Table 4. name @@v4@@ evaluating D. The global variables variables The fact that each ontological pattern involves more than one ICCD field, implies serious difficulties when it comes to translating one by one the fields of the same card. Table 3. Example of pattern serialization ICCD2OWL has to create a single pattern of instances that translate all of the fields involved, and cannot repeat the same pattern for each field involved. Moreover, more cards may B. The identity of the instances refer to the same entities involved in the same kind of Every time one proceeds to an object oriented representation, ontological pattern: what are the instances that can be reused it is necessary to choose a means of recognizing and and what are those that cannot? distinguishing the various instances of a certain class. Returning to our first example on the restoration data, we The simplest possible approach is that of assigning unique can notice that the five ICCD fields RSTD, RSTS, RSTE, progressive identifiers to the instances as they are created. But RSTN and RSTR all refer to the same ontology pattern this requires introducing some rules for detecting, identifiers represented in Fig. 3. This means, for example, that exactly apart, if two or more instances are equivalent, i.e. they refer to the same instance of the pattern Arch_Find-Restoration, the same entity. For example, before introducing whatever implemented for translating the first field RSTD, must be new instance, it would be necessary to verify whether no reused in the translation of all four of the other fields. instance representing the same entity is already recorded. Of course, there can be more than one ICCD card, In designing ICCD2OWL, a different policy was preferred. reporting different restorations of the same archaeological find Each instance whose scope is wider then one single card is in which the participants remain the same. In this case, the endowed with an identifier both unique and significant for the translations of the various restoration data will be made by entity that it represents. Every time two or more cards refer to means of patterns that share the instance of the endurants the same entity, it is ossible to create several instances all (Physical_Person, Juridical_Person, Archaeological having the same identifier: subsequently, the system is able to _Find_Person, etc.) but, for each restoration, each with its detect the duplications and to delete them. own time location at least, different sets of perdurants ( R e s t o r a t i o n and Participation) a n d different translate either ICCD cards into ReMuNaICCD instances or, Time_Region must be instantiated. viceversa, ReMuNaICCD instances into ICCD cards. In other words, if in translating RSTD the compiler A. Compiling ICCD creates two instances Participation_X1 and Participation_X2, then in translating the RSTS field the same instances Until now we have spoken about translating from ICCD to Participation_X1 and Participation_X2 must be reused. ReMuNaICCD. The following steps are performed as the Viceversa, to record further restorations of the same good, it ICCD file is sequentially read: will be necessary to implement new pairs of instances of 1. an ICCD field is read; Participation. 2. the appropriate variables are instantiated; In order to manage this distinction, the configuration file of 3. the ontological pattern is found that matches the field; ICCD2OWL is provided with global variables that are created 4. the elements of the pattern are processed sequentially; when certain precise fields are translated, and that remain at each step one of the following cases may occur: visible in all the subsequent processes, until the end of the a. the element is a subpattern that contains parameters, translation of all the correlated fields. then the parameters must be replaced by the actual values; b. the element is a triple, then its elements must be Trigger field RSTD instantiated using the current values of the variables; Pattern Restoration_@@RSTN@@ c. the element is a class, then the suitable instance of the class is instantiated. Name RX Of course, all the steps of the above process must be performed according to the rules coded in the configuration file previously introduced. TSK B. Compiling ReMuNaICCD ”RA” After the translation, the ontology is transferred into OWLIM [19] a high-performance semantic repository, where the reasoner computes the ontological inferred closure. In order to get both the original and the inferred data in a presentation format familiar to the customer, an OWL to ICCD compiler Table 5. Example of global variable declaration. can be of use. The process that converts the ontological information to When the compiler processes the first field concerning a the original ICCD format can be described in the following restoration, let us say RSTD, it creates also an instance of the steps. global variable RX, that contains the id of the instance of the For each instance of the ontology, and for each of its class Restoration. In this way, the information relevant to the properties: status of the translation remains available to the processes that 1. the ontological pattern is found that concerns the class translate the subsequent fields RSTS, RSTE, etc. of the instance and the property under consideration; An example of global variables codification is given in 2. the appropriate variables are instantiated; Table 5. 3. the elements of the pattern are processed sequentially; at each step one of the following cases may occur: E. The configuration file a. the element is the triple that evaluate variables, Summarizing, the configuration file of ICCD2OWL is then the value of the corresponding property must an XML document that contains the codifications of the be passed; following items: b. the element is a subpattern that makes references - a set of ontological patterns each of which, taken as a to variables, then the corresponding value must be whole, represents a not reducible conceptual structure; passed, and the process restart from point 2; - a set of mappings that represents conditional binding of 4 . the created pattern generates a query, that will be ICCD fields to variables inside the ontological performed on the ontology to retrieve the information patterns; about the property. - a set of mappings that bind the names used for the classes inside ReMuNaICCD, the symbolic names that VI. CONCLUSIONS represent them inside ICCD2OWL and the rule to Formal ontologies offer a promising approach to facilitate create a new instance; semantic information retrieval from heterogeneous document - a set of global variables and the corresponding repositories. instructions for use. The objective of our ontology was to restore a minimal logical structure to the numerous data that are found in the ICCD cards and whose relation with the original context V. AN OUTLINE OF THE COMPILER PROCESSES seems definitively lost. The formalism introduced above codifies the complete ReMuNaICCD models a “natural” infrastructure for the specifications that the compiler must apply in order to accommodation of the data. A first arrangement is determined by the taxonomy of the classes and the taxonomy of the properties, but the true logic of the model is contained in the [16] Patel-Schneider P. F., Hayes P., and Horrocks I. 2004, OWL web ontology language semantics and abstract syntax. W3C patterns of classes and properties that express the group games Recommendation, 10 Feb. 2004. http://www.w3.org/TR/owl-semantics. that the different entities play all together. [17] Masolo C., Borgo S., Gangemi A., Guarino N., Oltramari A. and Eventually, in trying to populate our ontology with data Schneider L. 2003. The WonderWeb Library of Foundational coming from the available sources, we realized that sensible Ontologies and the DOLCE ontology. WonderWeb Deliverable D18, translations of the traditional data models to ontology models Final Report (vr. 1.0, 31-12-2003) [18] CIDOC Conceptual Reference Model http://cidoc.ics.forth.gr/ were not at all trivial for two main reasons. [19] Kiryakov A., Ognyanov D., Manov D. 2005, OWLIM – a Pragmatic First, translations need pieces of information that could be Semantic Repository for OWL, in Proc. of International Workshop on acquired by the ontology patterns, but these patterns are not Scalable Semantic Web Knowledge Base Systems (SSWS 2005), explicitly represented inside the RDF/RDFS or OWL Galway, Ireland, November 6-10, 2005, LNCS 3729, pp. 668-684. ontology codifications. Second, the translations can be context dependent but none of the current ontology languages is able to express contextual dependence. Do these reasons mean that the current ontology languages are too poor for solving, without essential auxiliary tools, the problem of semantic interoperability? In this paper, we investigated the possibility of introducing supplementary formalizations that could be sufficient for the accomplishment of our specific task. Certainly, the technical details may be improved and the whole formalization may be better situated in the ambits of the most popular standards. REFERENCES [1] Pierobon Benoit R., Proto F., Aiello A., Brandi S., Mango Furnari M. 2005, Concettualizzazione e contestualizzazione dei beni culturali archeologici, «Archeologia e Calcolatori», 16, 321-339. [2] de Bruijn J., Polleres A., Lara R., and Fensel D. 2004, OWL Lite¯, in J. de Bruijn (ed.), Final draft d20.1v0.2, WSML Deliverble [ 3 ] McGuinness D. L. and van Harmelen F. 2004, (eds.), O W L W e b Ontology Language Overview, W3C Recommendation, 10 February 2004, http://www.w3.org/TR/2004/REC-owlfeatures-20040210/. [4] Volz L. 2000, Web Ontology Reasoning with Logic Databases. PhD thesis, AIFB, Karlsruhe. [5] Guarino N. and Welty C. 2000, Ontological analysis of taxonomic relationships. In Proceedings of ER-2000: The Conference on Conceptual Modeling. [6] Guarino N. 1998, Formal Ontology and Information Systems. In N. Guarino (ed.), Formal Ontology and Information Systems, Amsterdam, Nederlands, 3-15 [ 7 ] Guarino N. 1995, Formal Ontology, Conceptual Analysis and Knowledge Representation. «International Journal of Human and Computer Studies», 43(5/6), 625-640. [8] Gangemi A. 2005 Ontology Design Patterns for Semantic Web Content, in E. Motta and Y. Gil (eds.), Proceedings of the Fourth International Semantic Web Conference, LNCS 3729, 262 – 276. [ 9 ] Rector A.L., Rogers J. 2004, Patterns, Properties and Minimizing Commitment: Reconstruction of the GALEN Upper Ontology in OWL, in Core Ontologies Workshop (CORONT), Northampton, UK. [10] OEP 2004-5, Semantic Web Best Practices and Deployment Working Group, Task Force on Ontology Engineering Patterns. Description of work, archives, W3C Notes and recommendations available from http://www.w3.org/2001/sw/BestPractices/OEP/ (2004-5). [11] Gennari J., Musen M., Fergerson R., Grosso W., Crubézy M., Eriksson H., Noy N., and Tu S.2003, The evolution of Protégé-2000: An environment for knowledge-based systems development. «International Journal of Human-Computer Studies», 58(1), 89-123. [12] Knublauch H., Musen M. A., Rector A. L. 2004, Editing Description Logic Ontologies with Protégé OWL Plugin, Proceedings of the 2004 International Workshop on Description Logics, Whistler, British Columbia, Canada. [13] Broekstra J., Kampman A., van Harmelen F. 2002, Sesame: A Generic Architecture for Storing and Querying RDF and RDF Schema, in Proc. 1st ISWC. [14] Giordano M. and Mango Furnari M. 2006 Experimental Ontology Web Services. American Association for Artificial Intelligence. [15] Aiello A., Mango Furnari M., Massarotti A., Brandi S., Caputo V., and Barone V. 2006, An experimental ontology server for an information grid environment. «International Journal on Parallel Programming».