Answers are grounded in this paper and cite their sources.
Survey reveals challenges in task identification in low-context workplace messages, suggesting enhancements for future research.
The immediate embracing of instant messaging services in office environments has completely changed the way tasks, requests, and obligations are communicated. Although such platforms facilitate prompt coordination, they also raise the risk of tasks being missed or ignored, especially when the guidance is given in a very brief, casual, or indirect manner. Thus, automatic task identification in textual communication is, by now, an area of research with great interest. Most studies to date have been looking into rich-context sources like meeting transcripts, emails, and document-level text, where the surrounding text helps in understanding the meaning. Conversely, short workplace messages are usually cut off, thin, and indistinct; hence, they are quite a challenge for the task detection process. This survey elaborates on task and action detection in texts, particularly in short and low-context messages. We scheme previous research into thematic categories such as action-item detection, short-text intent recognition, classical machine learning approaches, and transformer-based models. By means of a critical analysis, we pinpoint major shortcomings in current research and emphasize the need for further study in the respective areas. The survey is designed to offer an organized view of the domain and to direct future research towards the development of powerful, low-context task detection systems adapted for real-world workplace communication.
No takes yet. Share an insight, caveat, or question.
UPADHYAY et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: