Enhancing crowdfunding financing speed through image-text integration strategies: Insights from Information Processing Theory and explainable machine learning methodology

Lending-based crowdfunding platforms provide entrepreneurs with a medium to present their projects' visual and textual information in order to secure funding. On these platforms, the ways in which entrepreneurs employ communication strategies that integrate images and text to describe their projects, attract investors, and expedite the fundraising process merit closer examination. In this context, the present study collected data from kiva.org and applied advanced image mining and text mining techniques to identify and quantify the features of image-text integration strategy (ITIS). Alongside other entrepreneur-controllable variables, a series of tree-based machine learning algorithms were utilized to develop a comparative and accurate model for predicting the financing speed of crowdfunding projects. Additionally, to ascertain which ITIS effectively enhance financing speed, an explanatory analysis of ITIS features was performed using Information Processing Theory (IPT) and explainable machine learning methods based on Ordinary Least Squares (OLS) and SHAP values. The key findings are as follows: (1) Different ITIS features exert varying degrees of influence on financing speed, with image-related features proving more crucial than text-related features, and content-related features holding greater importance compared to other feature types. (2) The specific effects of individual features on financing speed vary. For example, Image Content Richness (ICR), Image Sentiment (IS), Image Text Similarity (ITS), and Text Sentiment (TS) generally have positive impacts, while Image Aesthetic Value (IAV), Text Content Richness (TCR), and Text Readability (TR) may have negative effects. These findings provide entrepreneurs with an accurate predictive model for financing speed and offer valuable guidance for optimizing ITIS to enhance project performance.