Market-driven technology opportunity discovery by integrating trademark-based market signals and patent embeddings

This study proposes a market-driven technology opportunity discovery (TOD) framework that integrates trademark-based market signals with patent embeddings. Existing TOD studies have mainly relied on technology-oriented data, such as patents and scientific publications, to identify technological structures, trajectories, and vacant areas. Although these data are useful for understanding accumulated technological knowledge, they have limitations in capturing early market signals, product and service concepts, and firms’ commercialization intentions. To address this limitation, this study utilizes trademark designated goods descriptions as market-oriented textual data and links them with patent-based technological knowledge. The proposed framework consists of three main steps. First, trademark descriptions are transformed into sentence-level embeddings, and unsupervised anomaly detection algorithms are applied to identify trademark outliers that deviate from existing market patterns. Second, the identified trademark outliers are matched with semantically related patents in a shared embedding space to evaluate their technological relevance. Third, a patent-based generative topographic mapping (GTM) map is constructed, and trademark outliers are projected onto the patent-based technology space using the coordinates of their matched patents. Opportunity regions are then identified by comparing patent density and trademark signals at the cell level. Regions with strong trademark signals but low patent density are interpreted as market–technology imbalance areas and potential technology opportunities. The empirical results demonstrate that trademark data can complement patent-based analysis by revealing market-oriented signals that may not be captured from technological documents alone. This study contributes to technology opportunity discovery research by providing a systematic framework for linking early market signals with technological knowledge structures.