In the contemporary era, the landscape of innovation and entrepreneurship is dynamically evolving, fueled by a substantial surge in venture capital investments and the rapid expansion of the global startup ecosystem. This burgeoning growth not only highlights the vibrant nature of modern economies but also brings to the forefront the critical importance of identifying startups with high potential for success. As venture capital firms and investors seek to maximize their returns on investment, the ability to accurately assess and predict the future performance of these nascent companies becomes paramount. This dissertation delves into the heart of this challenge, aiming to refine and enhance the methodologies used in evaluating startup potential, thereby contributing valuable insights and tools to both academic scholars and industry practitioners. Existing methods for assessing startup potential have predominantly relied on static variables such as financial performance indicators, market size estimates, and competitive positioning. While these factors offer valuable insights, they fall short in capturing the dynamic and often unpredictable nature of startup growth and success. This raises several pertinent questions: How can we move beyond these traditional metrics to more accurately predict startup success? Furthermore, is it possible to develop more advanced tools that not only provide predictions but also facilitate a more interactive, dynamic evaluation process? These questions highlight the limitations of current approaches and pave the way for the innovative research presented in this dissertation, which seeks to explore these opportunities through the application of advanced data analytics and learning models. The dissertation is structured around three main chapters, each contributing to the overarching aim of developing a comprehensive framework for startup evaluation. The first chapter emphasizes the importance of mapping the interactions between various entities within the startup ecosystem, including companies, venture capital firms, and individuals. This interaction-centric view provides a foundational understanding of the complex interdependencies that influence startup success. Building on this foundation, the second chapter introduces an expanded interaction network and integrates company demographic features to improve the identification of high-potential startups. Additionally, this chapter explores the entrepreneurial homophily principle, which posits that startups with similar characteristics tend to cluster together, further supporting the theoretical underpinnings of the proposed methodologies. The third chapter represents a pioneering effort to leverage large language models (LLMs) for building an interactive, domain-centric tool aiming at dynamically evaluating startup potential. This novel application of LLMs opens up exciting possibilities for creating an interactive agent that can continuously update its assessments based on evolving data, offering a more fluid and responsive tool for venture capital decision-making. In summary, this dissertation marks a significant advancement in the field of startup evaluation by utilizing a diverse array of entrepreneurial data, combined with cutting-edge learning models. The research not only advances our theoretical understanding of startup dynamics but also offers practical tools for identifying startups with the highest potential for success. Through its comprehensive analysis and innovative methodologies, this work stands as a seminal contribution to the ongoing efforts to enhance the precision and relevance of startup potential assessment. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]