Framework for Automated Information Extraction from Meeting Minutes

Abstract

The construction industry is highly information dependent, and millions of textual documents are generated everyday within the industry. Meeting meetings are a very crucial part of the entire construction process, with daily, weekly, and monthly progress meetings playing an important role as a key means of correspondence amongst all involved stakeholders. Meeting minutes are generally analyzed and accessed manually, and this process leads to inefficiency in both time and money, as well as being quite prone to errors. Natural Language Processing (NLP) is a linguistic based approach that uses AI and rule-based programing to analyze text. Rule-based NLP was found to be more effective for this project. Over 40 samples of construction progress meeting minutes were obtained, and a standard format was defined. Our program was a Python script, using the samples of construction progress meeting minutes as input and putting it through processes like parsing, tokenization and POS- tagging using manually defined Information extraction rules. The output is in the form of an excel sheet which organizes all the extracted information under appropriate headings. This forms a database where all the analyzes construction progress meeting minutes are stored. This automated process is less prone to errors and delays in time, and is much more cost effective and user-friendly, as well as being more effective as analyzing a large amount of data.