Construction Trend Analysis Using GPT-3

Abstract

An Automated Keyword-Based Trend Analysis Service Such As Google Trends Is Helpful For Quickly Grasping Changes In Interest In A Particular Topic Over Time. However, The Current Keyword-Based Search Sometimes Returns A Nonsensical Result Due To Homonym Issues. This Study Proposes A Method That Analyzes Construction Trends Using The Semantic Search Function Of The Generative Pre-Trained Transformer 3 (GPT-3). GPT-3 Is A Deep-Learning-Based Language Prediction Model With A Capacity Of 175 Billion Machine Learning Parameters. Based On 11 Construction Technology Keywords, 1188 Score Data Are Extracted By Performing Queries And Calculations Over 11 Years, From 2010 To 2020, And From 9 Countries. Then, We Show The Difference Between The Obtained Data And The Search Results From Google Trends As Well As Check Whether Synonyms And Abbreviations Can Be Distinguished. The Method Suggested In This Study, Semantic Search Using GPT-3, Can Easily Retrieve The Latest Construction Trends.