Technology Opportunity Discovery of Cloud Computing Based on Latent Dirichlet Allocation Topic Model and Explainable Artificial Intelligence Model

This article describes the method of using Latent Dirichlet allocation (LDA) topic model and explainable artificial intelligence (XAI) model to analyze technological opportunities and utilize the results to develop research and development strategies. Technology opportunity discovery (TOD) is an analytical activity that explores the possibility of implementing technology topics base on sci-tech information and analysis model. As samples for analysis, 14340 patents related to cloud computing research applied from 2014 to 2023 are collected. Patent texts are processed by Latent Dirichlet allocation (LDA) topic model to extract technology topics and to reduce dimensions of classification models, inputs. Six kinds of machine learning models are used for classification of patent texts. The main driving features, that is, technology topics for patent authorization are extracted from the trained model with the best classification performance by Shapley additive explanations (SHAP) model. Technology opportunities are defined as these main driving features. Three kinds of technology opportunities about cloud computing are discovered.