Ontology-based economics knowledge sharing system

Highlights

• This paper argues the need for economics knowledge sharing.

• We demonstrate how economics knowledge sharing can be effectively achieved.

• Three search functions are implemented based on OEKSS.

• We propose instrumental variable recommendation algorithm (IVRA).

• IVRA provides a systematic way to find instrumental variables in OEKSS.

Abstract

The objective of this paper is to argue the need for economics knowledge sharing and to demonstrate that it can be achieved with Semantic Web technologies. To this end, we first designed an economics knowledge sharing ontology (EKSO) to describe economic domain knowledge. We then implemented an ontology-based economics knowledge sharing system (OEKSS) based on the EKSO and Semantic Web technologies. The OEKSS included three search functions – basic search, knowledge navigation, and instrumental variable recommendation – to demonstrate how we can use shared economics knowledge in future research. In particular, an instrumental variable recommendation is made based on an instrumental variable recommendation algorithm (IVRA), which is a systematic and efficient way to find instrumental variables through EKSO in limited experimental environments. Finally, the paper presents a case study for IVRA that illustrates the usefulness and significance of the algorithm.

Introduction

Economics knowledge consists of verified research and analyses of economic phenomena. This acquired knowledge is usually conveyed in digitalized forms, such as journal articles. To generate economics knowledge, economists follow three steps: modeling the target economic phenomenon; specifying the independent and dependent variables of the model and their relationships; and verifying the model by analyzing the corresponding data. This procedure is not so different from that of other experimental studies except in the third step.

Economists rarely perform experiments to gather data due to limited budgets, resources, space, and time. For instance, they cannot change a nation’s interest rate as part of an economic experiment. Therefore, in most cases, economic studies must be observational rather than experimental, and this makes it difficult to infer precise relationships among economic variables. Simply observing a correlation between two variables is not enough to conclude that there is a causal relationship between them (Wooldridge, 2001). Instead, economists need to conduct controlled experiments before drawing such a conclusion. This example highlights the need to share previous knowledge in economics. Although there have been few economic experiments, it is possible to understand the hidden ideas that coincide with our interests by sharing previous knowledge.

An instrumental variable (IV) of an economic model is a variable that is not included in the model, but it indirectly affects the dependent variable by being correlated with the independent variable (Hayashi, 2000). In economics, IVs are useful for identifying precise causal inferences when controlled experiments are impossible (Goldberger, 1972), but unfortunately, it is extremely difficult to find a suitable IV for a model. As a result, in most previous studies, the IV has been found for the model based solely on the author’s intuition. However, in this paper, we offer a systematic and an efficient way to find IVs by sharing economics knowledge.

To share economics knowledge in this article, Semantic Web technologies were used. The Semantic Web is an extension of the World Wide Web, whose content can be manipulated without human intervention (Berners-Lee, Hendler, & Lassila, 2001). One of the core technologies of the Semantic Web is ontology, which constitutes formal and consensual specifications of conceptualizations that provide a shared understanding of a domain (Gruber, 1993). In the Semantic Web, the content (i.e., semantic data) is expressed in a machine-interpretable format based on the concepts and relationships of the ontologies. Therefore, intelligent agents can understand the meaning of the content and share the domain knowledge (Fensel, 2001, Vesin et al., 2012, Yoo, 2012).

Using such Semantic Web technologies, we proposed an ontology-based economics knowledge sharing system (OEKSS) to demonstrate a way of sharing economics knowledge. To this end, an economics knowledge sharing ontology (EKSO) was designed to describe economic domain knowledge such as the metadata for the knowledge, concepts representing economic variables and their relationships. Economics knowledge can be collected from user participants. When a user registers economics knowledge pertaining to a certain economics paper, he or she can define not only the metadata for the paper, but also the relationships between the independent and dependent variables discussed in the paper. To assign the correct meaning to the variables, the user can also freely link the variables with their upper variables, without being constrained by a predefined set of variables. According to EKSO, the system transforms this knowledge into semantic data that the machine can interpret. To support economics knowledge sharing, basic search and knowledge navigation were implemented, and we also suggested a new method: the instrumental variable recommendation algorithm (IVRA) utilizing EKSO.

The rest of the paper is organized as follows. Section 2 reviews several systems to share specific domain knowledge closely related to this study. Section 3 presents the OEKSS architecture that we implemented on top of EKSO and illustrates how we can use the shared economics knowledge, including IV recommendations. In Section 4, the algorithm for the IV recommendations is examined more closely, and a case study is described using previous economics knowledge. Finally, Section 5 provides the conclusion to this paper, including possible limitations of our approach and potential direction for future researchers.