{
  "patent_number": "US 11842324",
  "country": "US",
  "title": "AI for Spotting Dam Problems in Inspection Reports",
  "original_title": "Method for extracting dam emergency event based on dual attention mechanism",
  "summary": "This patent describes an AI method using a \"dual attention mechanism\" to automatically find and organize information about dam emergencies from inspection reports, improving accuracy and reducing manual effort.",
  "what_it_does": "This method extracts dam emergency events from inspection reports using a sophisticated AI approach. First, it preprocesses the dam emergency reports by labeling and encoding sentences (Claim 1). Next, it builds a \"dependency graph\" that maps out how words relate to each other in sentences, considering both grammar and meaning, to identify dam emergency parameters (Claim 2). Then, it constructs a \"dual attention network\" which combines a Graph Transformer Attention Network (GTAN) to understand long-range word relationships and a standard attention network to capture key meanings, fusing their features to extract sentence-level event arguments (Claim 3). Finally, it fills in any missing details for an event at the document level by finding key sentences and looking for similar information in surrounding sentences using a \"twin neural network\" (Claim 4). For example, if a report mentions \"seepage in the foundation\" in one sentence and \"increased water levels\" in another, the system can link these to a single emergency event.",
  "what_it_does_not_cover": [
    "Does not cover event extraction for non-dam-related documents or general text analysis.",
    "Does not cover methods that do not use a dual attention mechanism combining a graph transformer network and an attention network.",
    "Does not cover predicting dam failures, only extracting reported events from existing text.",
    "Does not cover manual review processes for dam emergency events, as its purpose is automation.",
    "Does not cover event extraction without building a dependency graph based on sentence and semantic structure."
  ],
  "filed": "2022-10-14",
  "granted": "2023-12-12",
  "expires": "2042-10-14",
  "status": "active",
  "holder": "Hohai University HHU",
  "holder_url": "https://patentbrief.org/company/hohai-university-hhu",
  "inventors": [
    {
      "name": "Bingbing Nie",
      "url": "https://patentbrief.org/inventor/bingbing-nie"
    },
    {
      "name": "Chunrui Zhang",
      "url": "https://patentbrief.org/inventor/chunrui-zhang"
    },
    {
      "name": "Fugang ZHAO",
      "url": "https://patentbrief.org/inventor/fugang-zhao"
    },
    {
      "name": "Yingchi MAO",
      "url": "https://patentbrief.org/inventor/yingchi-mao"
    },
    {
      "name": "Hao Chen",
      "url": "https://patentbrief.org/inventor/hao-chen"
    },
    {
      "name": "Wenming Xie",
      "url": "https://patentbrief.org/inventor/wenming-xie"
    },
    {
      "name": "Fudong Chi",
      "url": "https://patentbrief.org/inventor/fudong-chi"
    },
    {
      "name": "Weiyong ZHAN",
      "url": "https://patentbrief.org/inventor/weiyong-zhan"
    },
    {
      "name": "Chunrui YANG",
      "url": "https://patentbrief.org/inventor/chunrui-yang"
    },
    {
      "name": "Haibin Xiao",
      "url": "https://patentbrief.org/inventor/haibin-xiao"
    },
    {
      "name": "Wei Sun",
      "url": "https://patentbrief.org/inventor/wei-sun"
    },
    {
      "name": "Han Fang",
      "url": "https://patentbrief.org/inventor/han-fang"
    },
    {
      "name": "Zhixiang Chen",
      "url": "https://patentbrief.org/inventor/zhixiang-chen"
    },
    {
      "name": "Xiaofeng Zhou",
      "url": "https://patentbrief.org/inventor/xiaofeng-zhou"
    },
    {
      "name": "Bin Tan",
      "url": "https://patentbrief.org/inventor/bin-tan"
    }
  ],
  "times_cited": 1,
  "tags": [
    "ai_ml",
    "software",
    "civil_engineering",
    "data_analytics"
  ],
  "abstract": "A method for extracting a dam emergency event based on a dual attention mechanism is provided. The method includes: performing data preprocessing, building a dependency graph, building a dual attention network, and filling a document-level argument. The performing data preprocessing includes labeling a dam emergency corpus and encoding sentences. Building a dependency graph includes assisting a model to mine a syntactic relation based on a dependency. Building a dual attention network includes weighing and fusing an attention network based on a graph transformer network (GTN) and capturing key semantic information in the sentence. Filling a document-level argument includes filling a document-level argument by detecting a key sentence and ordering a semantic similarity. The method introduces a dependency and overcomes the long-range dependency problem based on the dual attention mechanism, thus achieving high identification accuracy and reducing a lot of labor costs.",
  "url": "https://patentbrief.org/patent/us/11842324/method-for-extracting-dam-emergency-event-based-on-dual-attention-mechanism",
  "markdown_url": "https://patentbrief.org/patent/us/11842324/method-for-extracting-dam-emergency-event-based-on-dual-attention-mechanism/md",
  "google_patents_url": "https://patents.google.com/patent/US11842324",
  "relatedPatents": []
}