Knowledge Graph Construction for Patent Information in Offshore Wind Power Field

Facing the massive growth trend in patent text, high-quality patent analysis encounters significant challenges. To address the difficulties in processing and analyzing offshore wind power domain patent data, we constructed a patent knowledge graph based on data obtained from the internet. Initially, we retrieved relevant patents from patent databases, organized and cleaned the patent abstracts and descriptions to extract structured data, which was then transformed into 'entity-lationship-entity' triples. For unstructured data, we used dependency parsing and semantic tagging to extract knowledge. Subsequently, we applied knowledge fusion techniques to disambiguate extracted entities and relationships, removing redundant and erroneous information to ensure knowledge quality. Finally, the processed data was stored in the Neo4j graph database for graphical visualization. This research method and technology application provide strong support and tools for exploring offshore wind power domain patent knowledge in depth, offering new perspectives for related research.