Search for technological opportunities based on the patent array analysis

Currently companies still rely heavily on expert knowledge when searching for technological opportunities and choosing partners. Automation of search of patent holders who can be partners of the enterprises of the Volgograd region is considered. Companies not only from Russia, but also from China, India, and other friendly countries are considered as partners. The process of identifying partners is based on the similarity of solved technological problems extracted from patent documents, the patent holders of which are these enterprises. By analyzing the dependency trees extracted from the patent fields “Scope of the invention” and “Name of the invention” the “Problem-Solution” structures are formed. Based on the analysis of patent documents of potential partners “Problems-Solutions” that are not patented in Russia are identified. It causes a “technological vacuum” for enterprises of the Russian Federation. A method for identifying technological capabilities based on the analysis of the world patent array has been developed. The novelty of the method which provides a search for potential innovative technological areas for key enterprises of the Volgograd region lies in the use of deep learning technologies in relation to the analysis of natural language data of the world patent array. A software module has been developed to search for technological opportunities for enterprises of the Volgograd region based on the similarity of technological problems. The module is programmatically implemented in Python, the Yargy library from the Natasha project was used for semantic analysis of text fields of patents, Deep–translator was used for translating patent documents into Russian, Stanza library was used for building dependency trees, and RuWordNet library was used for selecting hyperonyms (to account for synonymous words). In order to identify technological opportunities for enterprises of the Volgograd region, 6 785 patents were analyzed.