SCIERC数据集解析

淩亂°似流年 2022-12-29 06:34 201阅读 0赞

### 文章目录 ###

*   *  数据集样例
     *  解析

[官方数据集地址][Link 1]

> Multi-Task Identification of Entities, Relations, and Coreference  
> for Scientific Knowledge Graph Construction

## 数据集样例 ##

*  六种实体类型
    
     *  Task, Method, Metric,Material, Other-ScientificTerm and Generic
 *  七种关系类型
    
     *  Compare, Part-of, Conjunction, Evaluate-for, Feature-of, Used-for, HyponymOf

{ 
      "clusters": [
        [
          [
            6,
            11
          ],
          [
            21,
            21
          ],
          [
            53,
            53
          ]
        ],
        [
          [
            15,
            16
          ],
          [
            69,
            69
          ],
          [
            94,
            94
          ]
        ],
        [
          [
            4,
            11
          ],
          [
            82,
            83
          ]
        ]
      ],
      "sentences": [
        [
          "This",
          "paper",
          "summarizes",
          "the",
          "formalism",
          "of",
          "Category",
          "Cooccurrence",
          "Restrictions",
          "-LRB-",
          "CCRs",
          "-RRB-",
          "and",
          "describes",
          "two",
          "parsing",
          "algorithms",
          "that",
          "interpret",
          "it",
          "."
        ],
        [
          "CCRs",
          "are",
          "Boolean",
          "conditions",
          "on",
          "the",
          "cooccurrence",
          "of",
          "categories",
          "in",
          "local",
          "trees",
          "which",
          "allow",
          "the",
          "statement",
          "of",
          "generalizations",
          "which",
          "can",
          "not",
          "be",
          "captured",
          "in",
          "other",
          "current",
          "syntax",
          "formalisms",
          "."
        ],
        [
          "The",
          "use",
          "of",
          "CCRs",
          "leads",
          "to",
          "syntactic",
          "descriptions",
          "formulated",
          "entirely",
          "with",
          "restrictive",
          "statements",
          "."
        ],
        [
          "The",
          "paper",
          "shows",
          "how",
          "conventional",
          "algorithms",
          "for",
          "the",
          "analysis",
          "of",
          "context",
          "free",
          "languages",
          "can",
          "be",
          "adapted",
          "to",
          "the",
          "CCR",
          "formalism",
          "."
        ],
        [
          "Special",
          "attention",
          "is",
          "given",
          "to",
          "the",
          "part",
          "of",
          "the",
          "parser",
          "that",
          "checks",
          "the",
          "fulfillment",
          "of",
          "logical",
          "well-formedness",
          "conditions",
          "on",
          "trees",
          "."
        ]
      ],
      "ner": [
        [
          [
            4,
            11,
            "Task"
          ],
          [
            6,
            11,
            "OtherScientificTerm"
          ],
          [
            15,
            16,
            "Method"
          ],
          [
            19,
            19,
            "Generic"
          ]
        ],
        [
          [
            21,
            21,
            "OtherScientificTerm"
          ],
          [
            23,
            24,
            "OtherScientificTerm"
          ],
          [
            31,
            32,
            "OtherScientificTerm"
          ],
          [
            36,
            38,
            "OtherScientificTerm"
          ],
          [
            47,
            48,
            "Method"
          ]
        ],
        [
          [
            53,
            53,
            "OtherScientificTerm"
          ],
          [
            56,
            57,
            "OtherScientificTerm"
          ],
          [
            61,
            62,
            "OtherScientificTerm"
          ]
        ],
        [
          [
            69,
            69,
            "Generic"
          ],
          [
            74,
            76,
            "Material"
          ],
          [
            82,
            83,
            "Task"
          ]
        ],
        [
          [
            94,
            94,
            "Method"
          ],
          [
            100,
            102,
            "OtherScientificTerm"
          ],
          [
            104,
            104,
            "OtherScientificTerm"
          ]
        ]
      ],
      "relations": [
        [
          [
            15,
            16,
            19,
            19,
            "USED-FOR"
          ]
        ],
        [],
        [
          [
            61,
            62,
            56,
            57,
            "FEATURE-OF"
          ]
        ],
        [
          [
            69,
            69,
            82,
            83,
            "USED-FOR"
          ],
          [
            74,
            76,
            69,
            69,
            "USED-FOR"
          ]
        ],
        [
          [
            100,
            102,
            104,
            104,
            "FEATURE-OF"
          ]
        ]
      ],
      "doc_key": "C88-1066"
    }

## 解析 ##

一条数据集包含：

1.  clusters 共指集群
    
        [
         [
           6,
           11
         ],
         [
           21,
           21
         ],
         [
           53,
           53
         ]
        ]
    
    比如最开头的这个集群，表示6到11个单词，第21个单词，第53个单词，指的是同一个实体。  
    sentence里的单词计数是从0开始。  
    这三对单词分别是：  
    “Category”, “Cooccurrence”, “Restrictions”, “-LRB-”, “CCRs”, “-RRB-”；  
    “CCRs”,  
    “CCRs”
2.  sentences 句子集群
    
    有几层外中括号，就分别对应sentences里的几个句子。
3.  ner 命名实体识别答案
    
    有几层外中括号，就分别对应sentences里的几个句子。  
    每个中括号里描述的当前句子里的命名实体识别
    
        [
            4,
            11,
            "Task"
        ]
    
    第一个数字代表实体的开头，第二个是实体的结尾。  
    “Task” 是实体类型
4.  relations 关系答案  
    有几层外中括号，就分别对应sentences里的几个句子。  
    每个中括号里描述的当前句子里的关系
    
        [
            69,
            69,
            82,
            83,
            "USED-FOR"
        ]
    
    四个数字分别是主体开头、主体结尾、客体开头、客体结尾。  
    “USED-FOR” 是关系类型
5.  doc\_key 文章编号

[Link 1]: http://nlp.cs.washington.edu/sciIE/