Ontology Evolution: A Case Study on Semantic Technology in the Media Domain

Ontology Evolution: A Case Study on Semantic Technology in the Media Domain

Christian Weiss (T-Systems Enterprise Services GmbH, Germany), Jon Atle Gulla (Norwegian University of Science and Technology, Norway), Jin Liu (Enterprise Services GmbH, Germany), Terje Brasethvik (Norwegian University of Science and Technology, Norway) and Felix Burkhardt (T-Systems Enterprise Services GmbH, Germany)
DOI: 10.4018/978-1-60566-650-1.ch006
OnDemand PDF Download:
No Current Special Offers


As semantic web technologies, including semantic search, have matured from visions to practical applications, this chapter describes a case study of (semi-) automatic construction and maintenance of ontologies and their applications to the media domain. A substantial amount of work has been done and will be done to integrate semantic search technologies into all kind of services where the underlying data is crucial for the success of automatic processing. Semantic search technologies should help both the user-to-machine and machine-to-machine communication to understand the meaning behind the data as well as to retrieve information according to user’s requests and needs. The crucial question is how to manage the semantic content (meaning) and how to deliver it in order to increase the value-chain of users’ benefits. Ontologies provide the basis for machine-based data understanding and ontology-based semantic search as they play a major role in allowing semantic access to data resources. However, the human effort needed for creating, maintaining and extending ontologies is often unreasonably high. In order to reduce effort for engineering and managing ontologies, the authors have developed a general framework for ontology learning from text. This framework has been applied in the media domain, in particular to video, music and later on to game search to offer an extended user experience in machineto machine as well as user-machine interaction.
Chapter Preview


Semantic technology is based mostly on automatic building of ontologies which is often referred to by the term “ontology learning”, originally introduced by Mädche, A. & Staab, S. (2001). It is normally described as the process of (semi-) automatically constructing ontologies on the basis of in-domain texts where a domain can represent a small part of a world-knowledge such as finance, media, or technical or cultural knowledge. The text data for learning is then taken out of this section of world-knowledge. The assumption is that the domain texts reflect the terminology that should go into the ontology, and that appropriate linguistic and statistical methods should be able to extract the appropriate concept candidates and their relationships from these texts. Semantic technology is an application-driven technology. In one of the most recent studies (Mills, D. 2008), more than 100 application categories have been examined, where semantic related technologies can be applied to. Generally speaking, semantics as a leading technology in the evolution of the Internet and the information society, should help to understand and manage the opinions freely expressed by people and make them not only understandable to humans but also to computers. This is the most essential focus in many applications, as machine-to-machine communication will play an increasingly important role in helping people to search, find and evaluate desired information. Most of the applications can be divided into two domains, the enterprise (B2B) domain and the private end consumer (B2C, or C2C) domain. In the enterprise domain, Enterprise Resource Planning (ERP), Enterprise Content Management (ECM), Supply Chain Management (SCM), Finance and Accounting (FA) as well as Project Lifecycle Management (PLM) are among others the applications deployed in mid-sized and large companies. In the consumer domain, topics like customer self-service, customer service automation, media entertainment, directories and information/product portals as well as mobile search profit strongly from semantic search technologies.

Semantically enabled search and management technologies have still been characterized as early market. The majority of current investments is rooted in R&D activities and rarely reaches operational deployment. More and more companies, however, are seriously considering the gradual introduction of the new technologies. Similar to classic motivations for investment, there are three basic elements for measuring the business value of semantic technologies:

  • Cost saving: This is to raise the efficiency. The purpose is to do the same job faster, cheaper and with fewer resources than previously.

  • Return on assets: This is to increase the effectiveness, doing a better job than you did before, and improving the productivity and performance.

  • Return on investment: This is to create new and /or value-added services by changing some existing business aspects and/or adding new strategic advantages.

Table 1 shows that semantic technologies can significantly reduce production and operating cost improve service quality and create new business.

Table 1.
Impact of semantic technologies on business value (adapted from Mills, D. 2004)
Impact of Semantic Technologies
Cost SavingReturn on AssetsReturn on Investment
20-80% less labor hours
20-90% less cycle time
30-60% less inventory levels
20-75% less operating costs
25-80% less set-up cost
20-85% less development cost
50-500% quality gain
2-50× productivity gain
2-10× greater number or complexity of concurrent projects, product reales, & units of work handled
2-25× increased return of assets
2-30× revenue growth
20-80% reduction in total cost ownership
3-12 month positive return on investment
3-300× positive ROI over 3-years

Complete Chapter List

Search this Book: