Meeting Intelligence: Task Rephrasing
Exploratory project to ‘decontextualize’ tasks and to-do items identified in meeting transcriptions, and rewrite each of them in a single sentence to appear in a separate to-do list. We build upon pretrained seq2seq transformer models, and modern techniques such as multitask learning, and touch upon unresolved issues such as automated metrics to evaluate NLG models.
In our teem, we build machine learning technology to detect tasks in emails, and explore the usage of this technology for the detection of tasks in meeting transcripts. In this exploratory project, we develop deep models for the decontextualization of tasks: given a task detected in the transcript, we identify its relevant context, and rephrase (rewrite) the task and its context into a coherent and self-contained sentence. This sentence may then appear as part of a separate to-do list.