Technology Opportunity Analysis for Creating Innovative Solutions: Applying Semi-supervised Topic Modelling on Patent Data

This paper introduces a technology opportunity analysis (TOA) framework designed to assist researchers in solving technological issues by leveraging insights from diverse domains. The methodology centers on the concept of functional approach and recombinant search, addressing technological issues of a specific domain by focusing on related functions realized in diverse reference fields. Beginning with the selection of a target technology and the collection of relevant data, the framework progresses to the extraction of reference fields and functions, utilizing patent classification codes and action-object structures of patent data. The zero-shot topic modeling module of BERTopic then clusters patent documents in a semi-supervised manner using reference field and function information, to provide technological insights of how different domains employ each function related to target technology. Next, the methodology leads to the evaluation process that prioritizes document clusters based on their potential value and technological feasibility. Finally, the document clusters with top scores are analyzed to extract detailed technological insights that can help solve the target technological issue. The proposed methodology was demonstrated with the case of miniaturization of wearable device. This study presents a data-driven TOA framework that fosters interdisciplinary collaboration, and extracts technology opportunities of high granularity using semi-supervised topic modeling technique.