A novel similarity-based recommendation for identifying potential customers in new markets using an inter-firm transaction network

In a dynamically evolving corporate landscape, it is essential to identify potential customers to ensure the sustainable growth of companies. In this context, potential customers can be identified by predicting the link that foreshadows future transactions between pairs of companies in an inter-firm transaction network. Similarity-based link prediction approaches are popular for predicting links, owing to their interpretability and scalability. However, existing similarity measures have proven inadequate for capturing intermarket similarities. This limitation restricts their applicability to scenarios in which businesses seek to enter new markets. To overcome this limitation, we propose a novel similarity score, designed to capture the similarities between firms in separate markets. The proposed similarity score is utilized to identify potential customers in new markets by leveraging transaction data along with essential firm attributes. We validate our approach through toy network experiments, visually demonstrating its ability to predict potential customers across different markets. Moreover, the proposed method consistently outperforms baseline approaches in terms of the Area Under the Curve (AUC), precision@k, and recall@k. These findings underscore the effectiveness of the proposed method as a valuable tool for businesses seeking to enter new markets.