A deep learning system to extract and structure key information from academic written texts

To further the speed and efficiency of knowledge acquisition from academic works, it would be useful to know, at quick glance, key information from the text. In academic works, key information would include named entities and their relationships, arguments, event descriptions, and topics contained in the text.
We propose an automated system to extract the above key information from PDF documents, to then be stored and structured. The stored and structured information would then be able to provide information to a user based on their requests. By building from our previous project, we focus now on the information in the text, rather than the structure. This interdisciplinary project combines natural language processing (NLP), computer vision, human-computer interaction (HCI), computational linguistics, text tokenization and preprocessing, entity extraction, text summarization, hierarchical multi-label classification, and textual logic relationships analysis.

Faculty Supervisor:

Seok-bum Ko;Roy Ka-Wei Lee

Student:

Partner:

Living Sky Technologies

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

University of Saskatchewan

Program: