{
  "patent_number": "US 9398104",
  "country": "US",
  "title": "How Facebook Uses User Feedback to Improve Search Results",
  "original_title": "Ranking test framework for search results on an online social network",
  "summary": "A method for improving search engine accuracy by letting users manually rate search results, then using those ratings to automatically adjust how the search algorithm ranks future results.",
  "what_it_does": "This patent describes a feedback loop for search algorithms within a social network. When a user performs a search, the system presents results that are personalized based on the user's connections in a social graph. The user then manually assigns scores to these results. The system uses these user-provided scores to calculate a 'discounted cumulative gain'—a mathematical metric that measures how well the search algorithm ordered the most relevant items at the top. The algorithm then updates its internal ranking logic to better prioritize similar results for future queries.",
  "what_it_does_not_cover": [
    "Does not cover search ranking systems that rely solely on automated click-through rates rather than explicit user-provided scores.",
    "Does not cover general search engines that operate without a social graph or user-connection data.",
    "Does not cover the specific machine learning models used to perform the ranking, only the method of using user feedback to modify those models."
  ],
  "filed": "2012-12-20",
  "granted": "2016-07-19",
  "expires": null,
  "status": "active",
  "holder": "Facebook Inc",
  "holder_url": "https://patentbrief.org/company/facebook-inc",
  "inventors": [
    {
      "name": "Sriram Sankar",
      "url": "https://patentbrief.org/inventor/sriram-sankar"
    },
    {
      "name": "Kihyuk Hong",
      "url": "https://patentbrief.org/inventor/kihyuk-hong"
    }
  ],
  "times_cited": 0,
  "tags": [
    "software",
    "ai_ml",
    "ecommerce"
  ],
  "abstract": "In one embodiment, a method includes accessing a social graph comprising a plurality of nodes and edges, receiving a set of scored results from a user that include results generated by a search algorithm in response to a query from the user and a score for each result, where each result corresponds to a node of the social graph, calculating a gain for each result based on the score of the result, and modifying the search algorithm based on the calculated gain.",
  "url": "https://patentbrief.org/patent/us/9398104/facebook-ads-manager",
  "markdown_url": "https://patentbrief.org/patent/us/9398104/facebook-ads-manager/md",
  "google_patents_url": "https://patents.google.com/patent/US9398104",
  "relatedPatents": []
}