Sign in Subscribe
Machine learning

Patent Awarded for System Improving Result Relevance in Search Engine Queries

Strategic Insights

— 1 min read

A patent has been granted for a system that improves result relevance in search engine queries by identifying realms applicable to search strings and matching them with high frequency words. The system, developed by Insight Direct USA Inc., uses a processor and computer-readable memory to receive search queries, parse them using natural language processing, and submit them to multiple search engines to generate results.

Key Takeaways:

  • The system improves result relevance by identifying realms applicable to search strings and matching them with high frequency words, reducing ambiguity and increasing accuracy.
  • The system uses a processor and computer-readable memory to receive search queries, parse them using natural language processing, and submit them to multiple search engines to generate results.
  • The system identifies realms applicable to the root domain of each search result by matching high frequency words contained within each result with high frequency words in the applicable realm.
  • The system sorts the search results into an ordered list based on whether the realm applicable to the root domain of each result matches the applicable realm of the search string.
  • The system allows users to specify multiple search engines to submit search queries to, increasing the likelihood of accurate results.
  • The system allows users to sort the ordered list based on search engine ranking, time of publication, and proximity of high frequency words to the mathematical center of the applicable realm.
  • The system uses unsupervised machine learning to adjust the realm threshold based on user feedback on the ordered list.

Statistics:

  • The patent was filed on April 3, 2024, and published online on June 3, 2025 (U.S. Patent Number 12321407).
  • The system processes search queries at a search portal, receiving input from users and generating results based on applicable realms and high frequency words.
  • The system uses a processor and computer-readable memory to perform the necessary processing and sorting of results.
  • The system can optionally sort results based on search engine ranking, time of publication, and proximity of high frequency words to the mathematical center of the applicable realm.

Sources:

  • U.S. Patent Number 12321407 (filed April 3, 2024, published June 3, 2025)
  • Kothari, Dhairya. Clustered metasearch. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(12321407)&db=USPAT&type=ids
  • Insight Direct USA Inc. (Chandler, Arizona, United States)
  • VerticalNews (Alexandria, Virginia, United States)

Read more