Ye, Deheng, et al. "APIReal: an API recognition and linking approach for online developer forums." Empirical Software Engineering 23.6 (2018): 3129-3160.

This paper addresses the challenge of extracting API mentions from natural language texts on social platforms like Stack Overflow and Twitter. The extraction process involves two key steps: API recognition, which distinguishes API mentions from other words, and API linking, which disambiguates these mentions to their specific fully qualified names. Given the informal and inconsistent nature of social content, these tasks are complicated by ambiguities such as common words used as API names, multiple APIs with similar names, and varying writing forms. The paper proposes a semi-supervised machine learning approach for API recognition, leveraging name synonyms and semantic context. For API linking, it introduces domain-specific heuristics such as mention-mention similarity, scope filtering, and mention-entry similarity. Evaluations are conducted on API mentions from Stack Overflow related to three libraries: Pandas, Numpy, and Matplotlib.