Distributed Construction of Ontologies Using Hozo Kouji Kozaki Eiichi Sunagawa Yoshinobu Kitamura Riichiro Mizoguchi The Institute of Scientific and Industrial Research (ISIR), Osaka University 8-1 Mihogaoka, Ibaraki, Osaka, 567-0047 Japan +81-6-6879-8416 {kozaki, sunagawa, kita, miz}@ei.sanken.osaka-u.ac.jp ABSTRACT Depended Ontology of B Ontology B Imports This paper discusses Hozo’s functionality for supporting a concept from Ontology A Ontology A distributed and collaborative construction of ontologies. In a Ontology B Dependent Ontology of A distributed environment, each ontology is revised asynchronously Imported concept by different developers. In such a situation, one of the key issues Another concepts is the maintenance of consistency among inter-dependent are defined by extending an ontologies. In order to realize consistent distributed development imported concept of ontologies, Hozo provides two functionalities: to manage the Ontology B dependencies between ontologies and to keep and restore depends on A consistencies of ontologies when they are changed. Constructs Ontology A Constructs Developer A Ontology B Keywords Developer B Figure 1. Distributed ontology development. Ontology, Distributed development, Dependency management dependent ontology of A. The authors call a development of 1. INTRODUCTION ontologies in such a manner distributed ontology development. Ontology is one of the key technologies to realize Semantic Web. The distributed ontology development applies to many situations: In the Semantic Web, ontologies are shared and reused by cooperative development, understanding the total picture of different developers in a distributed environment. To construct conceptual hierarchy of ontologies, reusing published ontologies large scale ontologies, it is necessary to collaborate with many and so on. developers. Therefore, distributed and collaborative construction In the distributed ontology development, developers construct of ontlogies is one of the most significant issues. In this paper, we multiple ontologies in cooperation among the developers. They discuss an ontology development system named Hozo to support a can reuse published components of other ontologies if possible. It construction of ontologies. We focus on providing a framework is a common way for ontology development to extend an existing for distributed development. Its main features include a ontology into a target-specific ontology. However, when dependency management of ontologies and a framework for developers construct ontologies in parallel or reuse ontology supporting to keep consistency of ontologies. Section 2 discusses which is under construction and thus unstable, consistency among a distributed ontology development we assume. Section 3 the ontologies is easily broken because they are revised summarizes Hozo and flow of distributed ontology development. asynchronously without notice. Furthermore, they are possibly The functionalities of Hozo to support such a development are updated without concerning whether other ontologies depend on described in section 4. In Section 5, discuss some related work. them or not and how those ontologies would be influenced by We conclude with a summary of some future works in Section 6, . their changes because authorities for maintenance of the ontologies are separated and distributed to each developer of them. 2. DISTRIBUTED CONSTRUCTION OF Therefore, when a developer changes his/her ontology, the change ONTOLOGY influences on dependent ontologies of it. In many cases, such a change may cause inconsistencies among the ontologies. We assume a situation where several ontologies are constructed separately in a distributed environment (and sometimes in parallel For consistent development of ontologies, a system should by different developers). In such a situation, some ontologies may manage dependencies among the ontologies and support their import concepts (classes) defined in other ontologies, and another developers to harmonize the ontologies. Practically, the system concept might be defined in the ontology by extending the should have at least two following functionalities: imported concepts (see Figure 1). And then, it means the ontology B which imports concepts from Ontology A depends on ontology 1) To manage ontologies with its dependencies on others. A. In this paper, we call ontologies which are depended by other 2) To provide a framework to keep and restore consistencies of ontology and those depend on others depended ontologies, and ontologies when they are changed. dependent ontologies, respectively. In Figure 1, Ontology A is the depended ontology of Ontology B, and Ontology B is the Based on this observation, the authors have investigated how a change of one ontology influences on others through its dependencies. And they have devised strategies for the change in Copyright is held by the author/owner(s). order to keep and restore the consistency of them [1]. They have WWW 2007, May 8--12, 2007, Banff, Canada. implemented a framework for ontology development in harmony among depended/dependent ontologies as an extension of Hozo (our ontology development tool) [2]. Shared Space Version Management Tracking Information Ontology Access Control Pane of changes (server system) Ontology which browsing （ modifying Manager the developer builds Dependency Ontology which Ontology Editor Management the developer refers to Upload Upload building Personal Space Download Ontology Server Upload Download (client system) ） Ontology/Model Management System Clients Reference / Install Developer Language Ontology (other agents) support Onto Studio Model Modification Modification Modification (a guide system for Editing Editing Editing ontology design) Developer Developer Developer Figure 3. A Conceptual Framework for Distributed Figure 2. Architecture of Hozo Ontology Development. defined in them. It implies that each developer has two kinds of 3. HOZO ontologies: ontologies which the developer builds and ontologies 3.1 Overview of Hozo which he/she refers to. We have developed an environment for building/using ontologies, Distributed ontology development proceeds in the repetition of the named Hozo, based on both of a fundamental consideration of an following steps; ontological theory and a methodology of building an ontology. The features of Hozo include 1) Supporting role representation [3, 1) A developer gets latest information on ontologies which he 4], 2) Visualization of ontologies in a well considered format, and builds or refers to from the ontology server. And he 3) Distributed development based on management of downloads them form shared space to personal space (client) dependencies between ontologies. Hozo is composed of Ontology through an ontology manager. If it is needed, he locks Editor, Onto-Studio (a guide system for ontology design), ontologies to avoid that someone updates those ontologies Ontology Server and Ontology Manager (see Figure 2). The while he is editing them. ontology editor provides a developer with a graphical interface, 2) The developer analyzes changes in the updated ontologies through which they can browse and modify an ontology locally. and evaluate whether the changes are influencing on The ontology server stores and manages ontologies under access consistency of the ontology which he is constructing with the control and version management. The developer can access and help of ontology manager. browse them through the ontology manager. Furthermore, the ontology editor of Hozo provides a user support module to 3) If the changes cause inconsistency in his ontology, the maintain consistencies of the dependencies among ontologies, developer modifies his ontology in order to keep and restore called Tracking Pane. Hozo’s native language is XML-based its consistency with the updated ontologies. Hozo helps this frame language and ontologies can be exported in OWL, and modification process by suggesting possible countermeasures RDF(S). It does not support native OWL, but it can import OWL for coping with each of the changes. partially 1 . The latest version of Hozo is published at the URL: http://www.hozo.jp. 4) After the modification, the developer starts editing his ontology as he needs. While editing the ontology, he can imports and use concepts from other ontologies which he 3.2 Flow of Distributed Ontology Development refers to. Then the dependency between his ontology and the Figure 3 shows a skeleton of our conceptual framework for referred ontology through the imported concepts is managed distributed ontology development. It consists of two parts in a by the ontology manager. server-client architecture. One is a shared space, where developers store ontologies to be opened to other developers. The other is a 5) After editing, the developer publishes his ontology by local space, where each developer builds and modifies each uploading it to the shared space. And then, he unlocks the ontology which he is responsible for. The developers cannot edit ontology if he allows other developers to edit it. the ontologies stored in the shared space directly. Under access Every developer goes over this process individually and in parallel, control and version management, they edit the personal copies of and then the whole target ontology evolves. As a result of their ontologies locally and upload them to the shared space when developing processes the whole target ontology is constructed in necessary. the shared space. In the distributed ontology development, a target ontology can be We suppose another collaborative development process such as regarded as a system of interrelated ontologies stored in the shared constructing a single ontology by many developers. Our space. They are constructed in cooperation among the developers. distributed ontology development also can support such process in Each developer constructs some of them under his responsibility2. the repetition of the following steps: Then, he may refer to other ontologies and import concepts 1) The developers share a target single ontology in the shared space. The ontology server manages versions of the ontology 1 The OWL import mechanism is under improvement. and accesses to it. 2 A same component ontology may be constructed by several 2) When a developer edits the target ontology, he locks the developers. ontology and downloads it to his personal space. In the ontology editor, imported concepts are represented with different color from other concepts, and the developer cannot modify4 them to keep consistencies of ontologies. 4.2 Harmonizing Interrelated Ontologies in their Conceptual Dependencies Ontologies Ontologies 4.2.1 Checking changes of depended ontologies in a client in a server Ontology Manager shows developers which ontology has been changed. To maintain the consistency of dependency, the Compare and Synchronize developer should get more information on, for example, that what ontologies in a client and a sever concepts/slots in the depended ontology have been changed and Show information which concepts in his ontology are influenced by the changes. of ontologies Hozo shows such information on the tracking pane and the browsing pane of its ontology editor. The tracking pane lists the changes in depended ontologies which Figure 4. A snapshot of Ontology Manger. influence on his ontology (see Figure 6). Those changes are classified in three types (deleted, modified and added), and their 3) If the ontology has been updated by another developer, he types are represented by icons. The changes are shown by nodes analysis the changes by comparing the ontology with old with icons in tree structure, and the developer can know which versions of it. The ontology manager supports him by concepts are influenced by the change through child nodes of the showing the changes and its influence. nodes. By clicking a node representing a concept, the selected 4) After the analysis, the developer edits the ontology. And then, concept in the ontology is pointed in the browsing pane of he uploads the edited ontology to shared space and unlocks it. ontology editor. In the browsing pane (see Figure 7), the ontology is visualized in 4. DISTRIBUTED CONSTRUCTION USING network structures, and the changed concepts are represented by same icons5 as tracking pane shows. When the developer selects a HOZO changed concept, the concepts influenced by the change are Here, the authors summarize how Hozo supports distributed and highlighted on the browsing pane. And then, if the change type of collaborative construction of ontologies. the selected concept is modification, the details are shown. 4.1 Dependency Management of Ontologies 4.2.2 Modifying the ontology to keep the consistency 4.1.1 Ontology Server and Ontology Manager To keep the consistency of the ontology, Hozo suggests possible The ontology manager (see Figure 4) acts as a bridge between the countermeasures for coping with each of the changes to the personal space (in a client) and the shared space which the developer. These countermeasures are devised through our ontology server provides. It carries out the following functions: investigation on conceptual dependencies of ontologies and the change type of Import Dialog 1) To show the latest information on the ontologies such as imported concepts [1]. “updated”, “locked by another developer” and so on. Selected In the beginning, Hozo 2) Access control to ontologies (lock and unlock) shows the developer 3) Version management of ontologies two major strategies: to accept the change 4) To search concepts defined in other ontologies and to reject it. The 5) Synchronize ontologies in client with ontologies in server latter implies to Concept depended by redefine the changed 4.1.2 Import concepts form other ontologies concept in his selected concept When the developer find reusable concepts defined in other ontology. For example, ontologies which are published in the server by other developers, if the change type is Selected concept he can import them to his ontology. The ontology manager modification of an supports him to import the concepts through Import Dialog of the imported concept, ontology manager (see Figure 5). The dialog shows concepts in acceptance of the the selected ontology by tree structure based on is-a relation of change corresponds to them, and the developer selects a concept which he wants to replace the imported import to his ontology. And then the system finds all the concepts concept with the Figure 5. Import Dialog. depended by the selected concept form its dependencies3, and it is imported with them. 4 The developer can use imported concepts to define another concept. For example, he can define sub classes of them. 5 On the browsing pane, sky blue nodes represent imported concepts from depended ontologies. Therefore, only sky blue 3 For example, super classes of the selected concept, and concepts nodes can have the icons because the changes appear only on which the selected concept refers to. the imported concepts in the distributed ontology development. Depended ontology Details of Modification Changed concepts and its type of change Concepts which the change Selected Concept influences Legends Legends Deleted Deleted Depended ontology Modified Modified Added Added Concepts which the change Figure 6. Tracking Pane. of selected concepts influences Figure 7. Representation of changes on Browsing Pane. modified one. And, if the change type is deletion of imported concept, the acceptance corresponds to deletion of the concept. among ontologies and its instance models based on our framework. The developer can apply these countermeasures by selecting it (2) Supporting native OWL without transformation of file format. through a pop up menu on the browsing pane. After applying (3) Functionality to deal with OWL imports which refer to OWL countermeasures, he edits his ontology for coping with the change files that are kept elsewhere. (4) Functionality for checking the if necessary. In that case, it is helpful to him that the system shows ontology with a reasoner. the concepts influenced by the change. Furthermore, if he needs advanced strategies, the system shows him all countermeasures6 7. ACKNOWLEDGMENTS with their details on a harmonizing pane. We are grateful to Mr. Mamoru Ohta for his support to implement our system. 5. Related Works PromptDiff includes a version-comparison algorithm and enables users to view the differences between the versions [5]. It takes 8. REFERENCES same approach with us, but it does not support distributed [1] E. Sunagawa, et al.: An Environment for Distributed development discussed section 2. DILIGENT [6] and ONKI [7] Ontology Development Based on Dependency Management, supports distributed development of ontology through shared Proc. of ISWC2003, pp. 453-468, 2003. space for ontologies in the same way with Hozo. But they do not [2] Kozaki K., et al.: Hozo: An Environment for Building/Using have functionalities to suggest countermeasures for coping with Ontologies Based on a Fundamental Consideration of "Role" each of the changes to the developer when depended ontologies and "Relationship", Proc. of EKAW2002, pp.213-218, are modified. KAON and ours focus on that changes in one Siguenza, Spain, 2002 ontology can cause inconsistencies in other dependent ontologies. And, in order to ensure their consistencies, they propose deriving [3] Sunagawa, E., et al. :Organizing Role-concepts in Ontology evolution strategies [8]. But it does not provides strategies which Development Environment: Hozo, Proc. of 2005 AAAI Fall reduce the influences against the changes although Hozo suggests Symposium on Roles, an interdisciplinary perspective, 2005 them (e.g. deletion of a concept can be canceled by redefining it in [4] Kozaki K., et al.: Fundamental Consideration of Role another ontology). The difference is caused by different treatment Concepts for Ontology Evaluation, Proc. of EON2006 of relationship between depended ontologies and dependent Edinburgh, United Kingdom, May 22, 2006 ontologies. [5] Noy, N., et al.: Tracking Changes during Ontology Evolution, 6. CONCLUSION AND FUTURE WORK Proc. of ISWC2004, Hiroshima, Japan, pp.259-273, 2004 In this paper, the authors discussed some functionalities of Hozo [6] Tempich, C., et al.: An Argumentation Ontology for to support distributed ontology construction. Harmonization of DIstributed, Loosely-controlled and evolvInG Engineering ontology is an essential issue especially in a distributed processes of oNTologies (DILIGENT), Proc. of ESWC2005, development. Our system contributes to resolving the issue based Greece, pp. 241-256, 2005 on management of dependencies between ontologies. The functionalities can support to construct a single ontology by many [7] Valo, A., Hyvonen, E. and Komurainen, V.: A Tool for developers collaboratively. Collaborative Ontology Development for the Semantic Web, in: Proc. of DC 2005, Madrid, Spain, 2005. As future work, the authors plan to enhance our system according to the following future plan: (1) Maintenance of consistency [8] Stojanovic, L., Maedche, A., Motik, B. and Stojanovic, N., User-driven Ontology Evolution Management, Proc. of 6 EKAW 2002, Madrid, Spain, pp. 285-300, 2002 We have not implemented some of advanced countermeasures yet. But, we suppose the two major strategies are enough for coping with the change in a lot of cases.