Exploring a technology ecology for technology opportunity discovery: A link prediction approach using heterogeneous knowledge graphs

Abstract

When a firm discovers and introduces new technology opportunities, it considers external context, such as technological changes, as well as internal context of technology capability and collaborators. These two contexts are separately considered in prior studies, despite their interaction. This study proposes a novel approach for technology opportunity discovery with a hybrid perspective using a technology ecology. It reflects technological contexts such as collaboration, competition, or technological association. We represent interactive relations between assignees, inventors, patents and technology areas via knowledge graphs, which are merged into a technology ecology. We focus on the relation between assignee and technology area nodes which indicates the assignee adopted technology opportunities related to the area. Further, to find the clues on new links, we analyze common neighboring nodes and position of the two nodes with network metrics. Next, we use them in a machine learning-based link prediction model, thereby identifying new technology areas likely to be linked to a firm of interest as technology opportunity candidates. They are evaluated in terms of the firm's internal context and external context and ranked via the TOPSIS method. The case study covering a technology-based firm in the biotechnology domain showed the applicability and feasibility of our approach.