Natural Language Processing Tools for Predictive Modeling of Advanced Trends in Formal Ontologies in Biomedical Sciences

Abstract

Natural language processing methods can be used to predict advanced application trends in formal ontologies. Formal ontologies help to formalize the characteristics of objects in various domains. As a result, machine learning programs identify patterns and relationships between these characteristics. The article describes an experiment based on machine learning methods in combination with text search methods. It involves the CatBoost algorithm for predictive modeling and clustering of lexical items. The vector models of the corresponding items reflect a trend in a particular domain of knowledge; proximity between them was calculated based on the idea of semantic distance. The experiment revealed four advanced areas for formal ontologies, i.e., genotype – phenotype; personalization; clustering algorithms, and collaborative task management. Each area that represented the predictable trends of development in this particular domain was provided with keywords. The article also contains a review of most popular scientific articles on these trends.

Keywords

• natural language processing
• formal ontologies
• bibliometric analysis
• predictive modeling research
• methods of machine learning
• vector models of lexical items
• clustering of lexical items
• catboost algorithm
• trends in intellectual development