Abstract The paper presents a methodology for the intelligent analysis of patent information, covering data collection, preprocessing, and both statistical and semantic analysis. Special emphasis is placed on the use of neural network models and Latent Dirichlet Allocation (LDA) to reveal latent structures in large-scale patent corpora.
Zrelova et al. (Mon,) studied this question.