Question and answer (Q&A) websites like Stack Overflow, AnswerHub, and Quora have become indispensable resources for developers, fostering collaboration and knowledge sharing in the software engineering community. However, the exponential growth of developer-generated content poses challenges in navigating these platforms effectively. Information Foraging Theory (IFT) offers a powerful framework for understanding information-seeking behaviors, yet its application to Q&A websites remains underexplored. We operationalize IFT for Stack Overflow, investigating value-cost factors and the dynamics of social and variation foraging among developers. Our semi-supervised machine learning models classify questions and answers based on their value-cost ratio, providing insights into foraging behaviors and facilitating the development of recommendation tools. By bridging theory and practice, our research enhances understanding of information-seeking behaviors on Q&A platforms and sheds light on novel foraging scenarios. Our models demonstrated an accuracy of 67.1% in classifying low-value questions and 83.7% in identifying high-cost questions. Additionally, they achieve an overall accuracy of 59.26% compared to actual value-cost estimations made by developers. The versatility of our approach extends beyond Stack Overflow to various domains, offering opportunities for comprehensive exploration of diverse foraging activities.