Weaving the Semantic Web the research community is working on publishing publicly available data sources as RDF data on the Web. To facilitate the adoption of Semantic Web technologies in corporate environments some issues on ontology engineering have to be addressed, e.g., support unexperienced employees to work collaboratively on ontologies. Although, existing methodologies structure well the process of ontology engineering, we miss an adequate tool support. We describe the Lekapidia case study and derive requirements for ontology engineering in a corporate environment. Furthermore, we present an extended ontology lifecycle integrating ontology engineering and ontology usage.
Within the past years the Semantic Web community has developed a
comprehensive set of standards and data formats to annotate semantically all kinds of
resources, e.g., documents and images. Currently, a main focus lies on
integrating publicly available data sources and publishing them as RDF on the Web,
e.g., linking open data [
To facilitate the adoption of Semantic Web technologies in a corporate environment some issues have to be addressed. Although ontology engineering methodologies might be well thought out, we miss an adequate tool support for unexperienced participants and the economic-driven needs of companies.
In Section 2 we present the results of the Lekapidia case study. A team of six people modeled collaboratively an ontology on desert recipes. Afterwards, this process was examined under the theoretic foundations of the DILIGENT methodology and simulated using a wiki-based tool for ontology engineering. We analyze the results of the case study in Section 3 and derive requirements on tools for modeling ontologies in a corporate environment. As a result we present an innovative two parts ontology lifecycle. Furthermore, we describe a corporate Semantic Web scenario, developed in cooperation with the Projektron GmbH: a semantic ticket system.
In this section we describe the Lekapidia case study which we use to evaluate
wiki-based ontology engineering empirically and to derive requirements for
ontology engineering in a corporate environment. First we outline the setting of
the case study and describe how the DILIGENT methodology was applied to
this scenario. Afterwards, we present our conclusions drawn from the case study
which refers to the special needs of corporate environments. In the Lekapidia
case study a team of six students were asked to develop collaboratively a
semantic wiki for desert recipes including a desert recipes ontology, while other four
people were working as the software engineers. The teams were free in choosing
the tools to build the ontology, to develop the application, to control the
collaborative work, and to produce a documentation. They used Protégé for modeling
tasks and a conventional MediaWiki for discussions.We used the case study for a
valuable proof-of-concept of the DILIGENT ontology engineering methodology
[
The participants of the Lekapidia project had no experience in ontology
engineering. Examining the activities of the working groups we discovered a lack of
communication. Considering DILIGENT, we discovered that the
argumentationbased approach has enabled the unexperienced users to discuss their design
decisions in an intuitive way. However, it does not support application-dependent
or scenario-oriented ontology engineering. Empirical studies such as [
Lekapidia does not allow any proposition about a long-running ontology engineering lifecycle, because the developed ontology was not deployed in a productive system. Even though, the structure of the project is close to real-world ontology development processes. The simulation with coefficientMakna allows to draw conclusions from a concrete methodological approach. We come to the conclusion that wiki-based approaches do not adequately support ontology engineering tasks. We identify a strong gap between the currently accepted ontology engineering approaches, e.g. DILIGENT, which suggest the applicability of wikibased ontology engineering, and the needs of ontology engineering in corporate contexts. Ontologies are commonly seen as an artifact without any applicationdependence. We suggest it as the outcome of a process which is concurrent while the ontology is in use and which is not finished after a decisive iteration step. Thus, appropriate methodologies and tools are needed, which respect this perspective. 3
The Lekapidia case study results a lack of adequate methodologies and tools respecting the agile character of ontology engineering. In the following we present new requirements for ontology lifecycles in a corporate environment. A key feature of our lifecycle is that it includes a cycle feeding back requirements on the ontology derived from its usage.
Corporate Semantic Web refers to the usage of Semantic Web technology in a corporate environment. In a project of the same name we focus besides others on ontology engineering in a collaborative environment to increase the effectiveness of this process. A main advantage over realizing the Semantic Web is the controlled environment in a company. That allows us to name the boundaries of the setting as follows: – Central allowance and control of the conceptualization – Existing rules and workflows for employees – Limited domain complexity – Trust in semantic annotations
A main part of ontology engineering is the evolution of an ontology over life time. Figure 1 depicts our approach towards a corporate ontology lifecycle. The outer circle describes the engineering process by ontology engineers and domain experts while the inner one describes the adaption of the ontologies driven by usage requirements.
The ontology engineering process (outer circle) starts with the creation/selection phase by collecting and model knowledge fragments which results in a prototype ontology. This ontology is validated against the objectives. At the intersection point between the engineering and the usage cycles the engineers decide if the ontology reached a state to be used (populated ) in the production system. If it does not meet the requirements or change requests arise from its usage the ontology engineers have to evaluate the current ontology. The evolution/forward engineering phase describes the task of changing the ontology to meet the new requirements. If an ontology has been populated to the production system then instances of concepts are generated by processing data and documents. The ontology is deployed. The feedback tracking phase is essential for adapting the 4 ontologies to new requirements arising from its usage. A new requirement arises if a user gives explicit feedback, e.g., by arguing about concepts and relationships, or a system monitor generates conclusions by tracking the user behavior. The collected feedback and requirements are analyzed (synchronization) and if inconsistencies are recognized between the user’s viewpoint and the ontology then ontology engineers start to adapt the ontology entering the outer cycle.
We transfer our lifecycle model into practice to support a feasible evaluation. In cooperation with the Projektron GmbH we evaluate the proposed ontology lifecycle in a real-world scenario: a semantic ticket system. Projektron uses and sells a ticket system which can be used to collect requests of customers, e.g., bug reports for a software. Although tickets are annotated with keywords and categories, similar tickets cannot be detected automatically. A main reason is the usage of synonym terms, e.g., differences in the terminology of the customer and the operating company of the ticket system. The difference in terminology may originate from the adaption of a software product to the terminology of the customer. For example, the customer uses “job” or “issue” instead of “task”. Having the information about similar tickets a software engineer could solve these tickets in a single run and, thus, save time.
Establishing an ontology lifecycle as described above helps to keep track of customer-specific changes in the terminology of a software product. Furthermore the ontology has to be adapted to the terminology of ticket submitters to be able to detect similar tickets in the system.
We name the following requirements for a semantic ticket system aiming at an integrative support for our lifecycle: 1. Expert design tools enable the operating company to develop valid and consistent ontologies. 2. (Semi-)Automatic knowledge acquisition performed by machine learning algorithms, amongst others, lessens the effort for the ontology engineers to develop valuable ontology prototypes, e.g., based on the common terminology of a customer. 3. (Semi-)Automatic knowledge retrieval lessens the additional work for the user to annotate relevant data at the run-time. 4. Lightweight extended communication platforms, e.g., forums or feedback forms, and the automatic recovery of user behavior feature the adaption of new requirements arising from ontology usage. 5. Alternative intuitive visualization, e.g., graph visualizer, provide an intuitive navigation for users of any level of experience and enable easy detection of similar concepts. 6. Interfaces for applications are necessary to allow a number of applications to integrate as much consistent ontologies as needed. 7. Ontology storage and versioning enable centrally administration and configuration of the interdependence, matching and alignment of the various coexisting ontologies.
We will extend and use these requirements in progress towards an architecture for a holistic corporate semantic web and implement a practical proof of concept which respects them. Thus, we do not just transfer Semantic Web technologies from web-scale to corporate-scale, but improve the foundations by innovative research results which start from another point of view, e.g. ontology engineering as a usage-oriented lifecycle. 4
Current lifecycle models for ontologies [
In order to find an applicable set of functional requirements for an integrative tool-support for ontology engineering in corporate environments, we used the results of the Lekapidia case study to discard wiki-based tools for this task. Based on ideas of the DILIGENT methodology and assumed characteristics of corporate settings, we constructed an innovative ontology lifecycle. The semantic ticketing use-case provides the basis for functional requirements which comply with the lifecycle integrative.
We expect the corporate ontology lifecycle to evolve towards a generic model, which enables companies to estimate the complexity and the chances of a transition from conventional information systems to ontology-based information systems. The approach will suite intra-corporate as well as inter-corporate settings.
Acknowledgement: This work has been partially supported by the ”InnoProfile-Corporate Semantic Web” project funded by the German Federal Ministry of Education and Research (BMBF).