I’m working on a text analysis project where I have short-to-medium-length reports (10–400 words) and a predefined list of tasks or actions (1 sentence each). The goal is to automatically detect which of these tasks are mentioned in each report.
Currently, I’m using a large language model to perform this matching. I’m considering adding short descriptions to each task so the model has more context, but I’m also open to alternative strategies, including embeddings, vector search, or any other NLP approaches.
I’m looking for ideas, best practices, or references for similar projects that handle task matching or action detection in short text reports.