Business model generation with link prediction

This study performs heterogeneous network link prediction to generate business model ideas. Company data were crawled from businessmodelideas, a platform that offers insights into corporate business models, to amass company descriptions and business model canvas information. Technology keywords the companies possess are extracted from the company description data using a technology keyword extraction tool. From the business model canvas data, keywords and phrases of revenue streams and value propositions are collected, embedded using SentenceBERT, and clustered based on semantic similarity through hierarchical clustering. A network is constructed based on the co-occurrence of technology, revenue stream, and value proposition keywords identified as the companies’ current business models. Link prediction is then applied to the heterogeneous network to ascertain potential business model archetypes that can be derived according to the newly formed edges. The findings of this study are anticipated to aid in the strategic planning and development of innovative business models.