An exploration and exploitation of value cocreation-based machine learning framework for automated idea screening

Idea screening in collaborative crowdsourcing communities poses significant challenges for firms. These challenges are primarily attributable to issues of prediction accuracy and information overload. The rapid expansion of idea pools generates a vast amount of data, making it difficult to effectively identify valuable ideas for new product development. This study introduces an interpretable framework for machine learning that integrates a novel exploration and exploitation perspective within the value cocreation model to enhance idea screening. The framework incorporates six theoretical dimensions of the exploration and exploitation of value cocreation (EEVC): the exploration and exploitation of digital resources, direct interactions, and ideas and their comments. Our evaluation reveals that the EEVC-based idea-screening system significantly outperforms the traditional 3Cs model in terms of prediction accuracy. SHAP value analysis further reveals that the exploration and exploitation of digital resources are the most influential predictors of idea implementation. The EEVC framework advances open innovation theory by clarifying how value cocreation dynamics influence idea implementation. Practically, it proposes a human–machine collaboration system that enhances expert decision-making for more effective idea selection.