Affiliations: Stanford University, Stanford, CA, USA. E-mail: [email protected] | American University of Beirut, Beirut, Lebanon and SRI International, Menlo Park, CA, USA. E-mail: [email protected]
Note: [] Corresponding author: Neil Yorke-Smith, American University of Beirut, Beirut 1107-2020, Lebanon and SRI International, Menlo Park, CA 94025, USA. E-mail: [email protected].
Abstract: Question–answer pairs extracted from email threads are valuable in constructing summaries of the content of the thread, as well as in providing data for semantic-based assistance with email. Previous work dedicated to extracting question–answer pairs from email threads considers only questions in interrogative form. We extend the scope of question and answer detection and pairing to encompass questions in imperative and declarative forms, and to operate at sentence-level fidelity. Building on prior work, our methods are based on learned models over a set of features that include the content, context, and structure of email threads. On multiple benchmark email corpora, we show that our methods balance precision and recall in extracting question–answer pairs, while maintaining a modest computation time.
Keywords: Email, question–answer pairing, information extraction, NLP
