Scholarlink

Patent application examination requires a lot of time and includes searching for similar documents registered before. Reducing the search part of the examination process will significantly improve the performance and throughput of a patent office. The paper provides research results of intellectual algorithms development and its application to the large data sets of patent data. The paper provides a comparison of search quality for different models, methods for improving search quality through metric learning, and the infrastructure used for the work.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intellectual Algorithms Based on Text Embeddings and Metric Learning for Searching Similar Documents in Patent Examination

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Alika Fazylova,
  • Vladimir Sennov,
  • Alexey Lukashin,
  • Alexander Gorbunov

摘要

Patent application examination requires a lot of time and includes searching for similar documents registered before. Reducing the search part of the examination process will significantly improve the performance and throughput of a patent office. The paper provides research results of intellectual algorithms development and its application to the large data sets of patent data. The paper provides a comparison of search quality for different models, methods for improving search quality through metric learning, and the infrastructure used for the work.