SmartCVC : A novel startup selection method for corporate venture capital

Corporate venture capital (CVC) has been growing rapidly in the past decades. As a critical first step for effective CVC investment, the selection of appropriate portfolio companies is challenging due to the large number of potential targets and the difficulty in evaluating the fit between CVC and different startups. In this study, we tackle this business problem through data science and develop an innovative prediction model, smartCVC, to recommend a short list of highly promising startups to a given CVC investor. To assess the strategic alignment and fit between a CVC investor and a startup, we adopt topic modeling techniques to analyze the business descriptions of both parties and construct a first-order topical similarity measure. We also construct two second-order topical similarity features from the existing CVC-portfolio investment network to characterize how similar a startup's past CVC investors are to a given CVC investor and how similar a CVC investor's past portfolios are to a given startup. Computational experiments on the CrunchBase dataset suggest that the proposed method achieves a significantly better performance than the baseline method.