Veröffentlichung

Titel:

Ontologies and Knowledge Graphs: A New Way to Represent and Communicate Values in Technology Design

AutorInnen:

Kathrin Bednar
Till Winkler

Kategorie:

Arbeitspapiere

erschienen in:

18th International Conference on the Ethical and Social Impacts of ICT, 2020

Abstract:

Technology is a mediator for human values, it moulds its use context, changes the perceptions and actions of people and creates new practices (Verbeek, 2008). In this view, negative effects do not solely emerge as a result of technology use, but are triggered by the affordances inherent in technology itself (van den Hoven, 2017). A promising pathway towards an ethically aligned design of technology is to incorporate human values such as privacy or autonomy during the design process. The most prominent approach to do so is Value sensitive Design (VSD; Friedman et al. 2006). VSD has inspired a number of related approaches and has developed 14 unique methods (Friedman, Hendry, & Borning, 2017). These methods produce a rich set of information through the identification of stakeholders, additional value sources, and the elicitation and analysis of values. However, not a single method focuses on the representation of values and the communication of value knowledge. We propose that ontology-building and the development of a knowledge base, approaches from the semantic web community, facilitate the representation and communication of values in an objective,
valid, complete, and transparent way. From an engineering perspective, the translation of value knowledge into formally defined terms makes the fuzzy concept of values more tangible and can therefore help to bridge the gap between disciplines. In this paper, we develop a value ontology and a knowledge base for a product case study as a proof of concept. The semantic web is an extension of the current web that aims at representing the semantic dimension of information in a formal, machine-readable way. Within the vision of the semantic web, ontologies play a key role in providing formally defined terms. An ontology is an explicit description of concepts (or classes), their properties, and restrictions, within an area of interest. In our case, the area of interest comprises human values for a specific technology context and their relations among each other as well as with the stakeholders. Ontologies can be used flexibly to define concepts and relations, but also allow the definition of constraints (e.g. a value being relevant only for the affected stakeholders is a constraint). Once an ontology is “filled” with individual instances (i.e. specific values for a specific technology context), a knowledge base is formed. The information contained in the knowledge base can be visually represented as graphs. A graph connects nodes (concepts) and thus represents their relation. An aggregated knowledge graph in turn allows the derivation of new knowledge, for example through querying the underlying information structure. Among others, building an ontology helps to share information among people, to analyze and reuse knowledge, and to make assumptions explicit (Noy & McGuinness, 2001). Building ontologies to represent values can not only improve the
communication of value knowledge, but also allows building on existing knowledge and coming up with new knowledge through additional ways of value analysis. Value knowledge should be represented and communicated in a transparent and traceable way throughout the whole design and development process. Table 1 summarizes requirements important for a high quality method to represent and communicate value knowledge. We consider these requirements as currently underrepresented in VSD methods. In the extended version of this paper, we discuss to which degree the approach we present can live up to these requirements.

Download: