Lecturers typically conduct surveys in an effort to accumulate information from a predefined group of scholars to achieve insights into matters of curiosity. When analyzing surveys with open-ended textual responses, this can be very time-consuming, labor-intensive, and tough to manually course of all of the responses into an insightful and complete report. Within the evaluation step, historically, the instructor has to learn every of the responses and resolve on tips on how to group them in an effort to extract insightful info. Despite the fact that it’s attainable to group the responses solely utilizing sure key phrases, such an strategy can be restricted because it not solely fails to account for embedded contexts but additionally can not detect polysemous phrases or phrases and semantics that aren’t expressible in single phrases. On this work, we current a novel end-to-end context-aware framework that extracts, aggregates, and abbreviates embedded semantic patterns in open-response survey information. Our framework depends on a pre-trained pure language mannequin in an effort to encode the textual information into semantic vectors. The encoded vectors then get clustered both into an optimally tuned variety of teams or right into a set of teams with pre-specified titles. Within the former case, the clusters are then additional analyzed to extract a consultant set of key phrases or abstract sentences that function the labels of the clusters. In our framework, for the designated clusters, we lastly present context-aware wordclouds that reveal the semantically outstanding key phrases inside every group. Honoring person privateness, we now have efficiently constructed the on-device implementation of our framework appropriate for real-time evaluation on cell gadgets and have examined it on an artificial dataset. Our framework reduces the prices at-scale by automating the method of extracting probably the most insightful info items from survey information.