Research SummaryExtant research in neuroscience suggests that human perception is multimodal in nature—we model the world integrating diverse data sources such as sound, images, taste, and smell. Working in a dynamic environment, entrepreneurs are expected to draw on multimodal inputs in their decision making. However, extant research in entrepreneurship has largely focused on how entrepreneurs or investors develop insights from data in a single mode. A few studies that have used a multimodal approach either simplify the multimodal data (MMD) into a few constructs or manually analyze the data without fully utilizing their potential. Such oversimplification limits the insights that can be gained from MMD. In this paper, we offer a framework to guide researchers to analyze and integrate MMD, capturing various cues embedded in the entrepreneurial process. We illustrate how applying machine learning algorithms to MMD can engender a robust, reliable, and scalable approach for researchers to effectively capture the elusive yet critical aspects of entrepreneurial phenomena. We also curate a set of data and algorithm resources for researchers interested in leveraging MMD in their studies.Managerial SummaryEntrepreneurs operate in fast‐paced and complex environments where success often relies on the ability to make sense of diverse and rich information, which ranges from explicit observations (e.g., what they see and hear) to more subtle contextual cues. Yet, most entrepreneurship research focuses on analyzing data in a single mode, such as only texts or numbers. Our research highlights the importance of embracing multimodal data (MMD) that combines various formats like audio, image, video, and text, to better understand and explain entrepreneurial decision‐making. We introduce a practical framework and a set of machine learning techniques that help managers and researchers alike harness the richness of multimodal data. Rather than simplifying or manually analyzing multimodal data, our approach allows for scalable, systematic, and reliable insights into the entrepreneurial process. For practitioners, this means better tools for evaluating pitches, tracking team dynamics, or sensing market trends in real time. To support adoption, we also provide a curated set of MMD sources and algorithms that organizations can leverage to make more informed strategic decisions.

中文翻译：

---

集成多模态数据和机器学习进行创业研究

研究摘要神经科学的现有研究表明，人类感知本质上是多模态的——我们整合了声音、图像、味觉和嗅觉等各种数据源对世界进行建模。在动态环境中工作，企业家需要在决策中利用多式联运的意见。然而，现有的创业研究主要集中在企业家或投资者如何以单一模式从数据中获得见解。一些使用多模态方法的研究要么将多模态数据 （MMD） 简化为几个结构，要么手动分析数据而没有充分利用其潜力。这种过度简化限制了可以从 MMD 获得的见解。在本文中，我们提供了一个框架来指导研究人员分析和整合 MMD，捕捉嵌入创业过程中的各种线索。我们说明了将机器学习算法应用于 MMD 如何为研究人员产生一种强大、可靠和可扩展的方法，以有效地捕捉创业现象中难以捉摸但关键的方面。我们还为有兴趣在研究中利用 MMD 的研究人员策划了一组数据和算法资源。管理总结企业家在快节奏和复杂的环境中运作，成功通常取决于理解多样化和丰富信息的能力，这些信息的范围从明确的观察（例如，他们的所见所闻）到更微妙的上下文线索。然而，大多数创业研究都侧重于以单一模式分析数据，例如仅分析文本或数字。我们的研究强调了采用多模态数据 （MMD） 的重要性，这些数据结合了音频、图像、视频和文本等各种格式，以更好地理解和解释创业决策。 我们介绍了一个实用框架和一组机器学习技术，可帮助管理人员和研究人员利用丰富的多模态数据。我们的方法不是简化或手动分析多模式数据，而是允许对创业过程进行可扩展、系统和可靠的洞察。对于从业者来说，这意味着更好的工具来评估推销、跟踪团队动态或实时感知市场趋势。为了支持采用，我们还提供了一组精选的 MMD 源和算法，组织可以利用这些资源和算法来做出更明智的战略决策。