Azure AI Search에서 더 이상 사용되지 않는 인지 기술

이 문서에서는 사용되지 않는 것으로 간주되는 기술을 설명합니다(더 이상 사용되지 않음). 내용에 대해 다음 가이드를 사용합니다.

  • 기술 이름: 더 이상 사용되지 않는 기술의 이름입니다. 특성에 @odata.type 매핑합니다.
  • 마지막 사용 가능한 API 버전: 더 이상 사용되지 않는 해당 기술을 포함하는 기술 세트를 만들거나 업데이트할 수 있는 Azure AI Search 공용 API의 마지막 버전입니다. 이러한 기술을 사용하는 연결된 기술 세트가 있는 인덱서는 "지원 종료" 날짜까지 향후 API 버전에서도 계속 실행되며, 이때 실패하기 시작합니다.
  • 지원 종료: 해당 기술이 지원되지 않는 것으로 간주되고 작동이 중지되는 날입니다. 이전에 만든 기술 세트는 계속 작동해야 하지만 사용자는 사용되지 않는 기술에서 벗어나 마이그레이션하는 것이 좋습니다.
  • 권장 사항: 지원되는 기술을 사용하기 위한 마이그레이션 경로입니다. 사용자는 권장 사항을 따라 지원을 계속 받는 것이 좋습니다.

Microsoft.Skills.Text.EntityRecognitionSkill(엔터티 인식 인식 기술(v2))을 사용하는 경우 이 문서는 일반적으로 사용할 수 있으며 새로운 기능을 소개하는 Microsoft.Skills.Text.V3.EntityRecognitionSkill을 사용하도록 기술 세트를 업그레이드하는 데 도움이 됩니다.

Microsoft.Skills.Text.SentimentSkill(감정 인지 기술(v2))을 사용하는 경우 이 문서는 일반적으로 사용할 수 있고 새로운 기능을 소개하는 Microsoft.Skills.Text.V3.SentimentSkill을 사용하도록 기술 세트를 업그레이드하는 데 도움이 됩니다.

Microsoft.Skills.Text.NamedEntityRecognitionSkill(명명된 엔터티 인식 인식 기술(v2))을 사용하는 경우 이 문서는 일반적으로 사용할 수 있고 새로운 기능을 도입하는 Microsoft.Skills.Text.V3.EntityRecognitionSkill을 사용하도록 기술 세트를 업그레이드하는 데 도움이 됩니다.

Microsoft.Skills.Text.EntityRecognitionSkill

마지막 사용 가능한 API 버전

2021-04-30-Preview

지원 종료

2024년 8월 31일 토요일

권장 사항

대신 Microsoft.Skills.Text.V3.EntityRecognitionSkill을 사용합니다. EntityRecognitionSkill의 대부분의 기능을 더 높은 품질로 제공합니다. 또한 복잡한 출력 필드에 다양한 정보가 있습니다.

Microsoft.Skills.Text.V3.EntityRecognitionSkill로 마이그레이션하려면 기술 정의를 다음 중 하나 이상 변경합니다. 기술 세트 업데이트 API사용하여 기술 정의를 업데이트할 수 있습니다.

  1. (필수)@odata.type"#Microsoft.Skills.Text.EntityRecognitionSkill"에서 "#Microsoft.Skills.Text.V3.EntityRecognitionSkill"로 변경

  2. (선택 사항) 새 기술이 알려진 형식의 엔터티만 반환하므로 매개 변수가 includeTypelessEntities 더 이상 지원되지 않으므로 이전 기술 정의에서 참조한 경우 이제 제거해야 합니다.

  3. (선택 사항) 출력을 사용하는 namedEntities 경우 속성 이름을 약간 변경합니다.

    1. value 이름이 로 바뀝니다. text
    2. confidence 이름이 로 바뀝니다. confidenceScore

    정확히 동일한 속성 이름을 생성해야 하는 경우 ShaperSkill추가하여 필요한 이름으로 출력의 모양을 바꿉니다. 예를 들어 이 ShaperSkill은 속성 이름을 이전 값으로 바꿉니다.

    {
        "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
        "name": "NamedEntitiesShaper",
        "description": "NamedEntitiesShaper",
        "context": "/document/namedEntitiesV3",
        "inputs": [
            {
                "name": "old_format",
                "sourceContext": "/document/namedEntitiesV3/*",
                "inputs": [
                    {
                        "name": "value",
                        "source": "/document/namedEntitiesV3/*/text"
                    },
                    {
                        "name": "offset",
                        "source": "/document/namedEntitiesV3/*/offset"
                    },
                    {
                        "name": "category",
                        "source": "/document/namedEntitiesV3/*/category"
                    },
                    {
                        "name": "confidence",
                        "source": "/document/namedEntitiesV3/*/confidenceScore"
                    }
                ]
            }
        ],
        "outputs": [
            {
                "name": "output",
                "targetName": "namedEntities"
            }
        ]
    }
    
  4. (선택 사항) 출력을 entities 사용하여 엔터티를 잘 알려진 엔터티에 연결하는 경우 이 기능은 이제 새로운 기술인 Microsoft.Skills.Text.V3.EntityLinkingSkill입니다. 연결된 엔터티를 생성하려면 엔터티 연결 기술을 기술 세트에 추가합니다. 또한 EntityRecognitionSkill과 새 EntityLinkingSkill 사이에 entities 출력의 속성 이름이 약간 변경되었습니다.

    1. wikipediaId 이름이 로 바뀝니다. id
    2. wikipediaLanguage 이름이 로 바뀝니다. language
    3. wikipediaUrl 이름이 로 바뀝니다. url
    4. typesubtype 속성은 더 이상 반환되지 않습니다.

    정확히 동일한 속성 이름을 생성해야 하는 경우 ShaperSkill추가하여 필요한 이름으로 출력의 모양을 바꿉니다. 예를 들어 이 ShaperSkill은 속성 이름을 이전 값으로 바꿉니다.

    {
        "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
        "name": "LinkedEntitiesShaper",
        "description": "LinkedEntitiesShaper",
        "context": "/document/linkedEntitiesV3",
        "inputs": [
            {
                "name": "old_format",
                "sourceContext": "/document/linkedEntitiesV3/*",
                "inputs": [
                    {
                        "name": "name",
                        "source": "/document/linkedEntitiesV3/*/name"
                    },
                    {
                        "name": "wikipediaId",
                        "source": "/document/linkedEntitiesV3/*/id"
                    },
                    {
                        "name": "wikipediaLanguage",
                        "source": "/document/linkedEntitiesV3/*/language"
                    },
                    {
                        "name": "wikipediaUrl",
                        "source": "/document/linkedEntitiesV3/*/url"
                    },
                    {
                        "name": "bingId",
                        "source": "/document/linkedEntitiesV3/*/bingId"
                    },
                    {
                        "name": "matches",
                        "source": "/document/linkedEntitiesV3/*/matches"
                    }
                ]
            }
        ],
        "outputs": [
            {
                "name": "output",
                "targetName": "entities"
            }
        ]
    }
    
  5. (선택 사항) 명시적으로 지정 categoriesEntityRecognitionSkill V3 하지 않으면 에서 지원EntityRecognitionSkill한 범주 외에 다른 유형의 범주를 반환할 수 있습니다. 이 동작이 바람직하지 않은 경우 명시적으로 categories 매개 변수를 ["Person", "Location", "Organization", "Quantity", "Datetime", "URL", "Email"]으로 설정해야 합니다.

    샘플 마이그레이션 정의

    • 간단한 마이그레이션

      (이전) EntityRecognition 기술 정의

      {   
          "@odata.type": "#Microsoft.Skills.Text.EntityRecognitionSkill",
          "categories": [ "Person" ],
          "defaultLanguageCode": "en",
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              }
          ]
      }
      

      (이후) EntityRecognition 기술 V3 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.V3.EntityRecognitionSkill",
          "categories": [ "Person" ],
          "defaultLanguageCode": "en",
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              }
          ]
      }
      
    • 복잡한 마이그레이션

      (이전) EntityRecognition 기술 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.EntityRecognitionSkill",
          "categories": [ "Person", "Location", "Organization" ],
          "defaultLanguageCode": "en",
          "minimumPrecision": 0.1,
          "includeTypelessEntities": true,
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              },
              {
                  "name": "namedEntities",
                  "targetName": "namedEntities"
              },
              {
                  "name": "entities",
                  "targetName": "entities"
              }
          ]
      }
      

      (이후) EntityRecognition 기술 V3 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.V3.EntityRecognitionSkill",
          "categories": [ "Person", "Location", "Organization" ],
          "defaultLanguageCode": "en",
          "minimumPrecision": 0.1,
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              },
              {
                  "name": "namedEntities",
                  "targetName": "namedEntitiesV3"
              }
          ]
      },
      {
          "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
          "name": "NamedEntitiesShaper",
          "description": "NamedEntitiesShaper",
          "context": "/document/namedEntitiesV3",
          "inputs": [
              {
                  "name": "old_format",
                  "sourceContext": "/document/namedEntitiesV3/*",
                  "inputs": [
                      {
                          "name": "value",
                          "source": "/document/namedEntitiesV3/*/text"
                      },
                      {
                          "name": "offset",
                          "source": "/document/namedEntitiesV3/*/offset"
                      },
                      {
                          "name": "category",
                          "source": "/document/namedEntitiesV3/*/category"
                      },
                      {
                          "name": "confidence",
                          "source": "/document/namedEntitiesV3/*/confidenceScore"
                      }
                  ]
              }
          ],
          "outputs": [
              {
                  "name": "output",
                  "targetName": "namedEntities"
              }
          ]
      },
      {
          "@odata.type": "#Microsoft.Skills.Text.V3.EntityLinkingSkill",
          "defaultLanguageCode": "en",
          "minimumPrecision": 0.1,
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "entities",
                  "targetName": "linkedEntities"
              }
          ]
      },
      {
          "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
          "name": "LinkedEntitiesShaper",
          "description": "LinkedEntitiesShaper",
          "context": "/document/linkedEntitiesV3",
          "inputs": [
              {
                  "name": "old_format",
                  "sourceContext": "/document/linkedEntitiesV3/*",
                  "inputs": [
                      {
                          "name": "name",
                          "source": "/document/linkedEntitiesV3/*/name"
                      },
                      {
                          "name": "wikipediaId",
                          "source": "/document/linkedEntitiesV3/*/id"
                      },
                      {
                          "name": "wikipediaLanguage",
                          "source": "/document/linkedEntitiesV3/*/language"
                      },
                      {
                          "name": "wikipediaUrl",
                          "source": "/document/linkedEntitiesV3/*/url"
                      },
                      {
                          "name": "bingId",
                          "source": "/document/linkedEntitiesV3/*/bingId"
                      },
                      {
                          "name": "matches",
                          "source": "/document/linkedEntitiesV3/*/matches"
                      }
                  ]
              }
          ],
          "outputs": [
              {
                  "name": "output",
                  "targetName": "entities"
              }
          ]
      }
      

Microsoft.Skills.Text.SentimentSkill

마지막 사용 가능한 API 버전

2021-04-30-Preview

지원 종료

2024년 8월 31일 토요일

권장 사항

대신 Microsoft.Skills.Text.V3.SentimentSkill을 사용합니다. 개선된 모델을 제공하며. 의견 마이닝 또는 측면 기반 감정을 추가하는 옵션을 포함합니다. 기술이 훨씬 더 복잡해지면 출력도 매우 다릅니다.

Microsoft.Skills.Text.V3.SentimentSkill로 마이그레이션하려면 다음 중 하나 이상을 기술 정의에 변경합니다. 기술 세트 업데이트 API사용하여 기술 정의를 업데이트할 수 있습니다.

참고 항목

Sentiment Skill V3의 기술 출력은 SentimentSkill을 기반으로 하는 인덱스 정의와 호환되지 않습니다. 감정 기술을 새 버전으로 바꾸려면 인덱스 정의, 기술 세트(나중에 기술 입력 및/또는 지식 저장소 예측) 및 인덱서 출력 필드 매핑을 변경해야 합니다.

  1. (필수)@odata.type"#Microsoft.Skills.Text.SentimentSkill"에서 "#Microsoft.Skills.Text.V3.SentimentSkill"로 변경

  2. (필수) 감정 기술 V3는 전체 텍스트의 각 문장에 대해 전체 텍스트 및 negative 동일한 점수에 대한 점수를 neutral제공하는 반면, 이전 SentimentSkill은 전체 텍스트에 대해 0.0(음수)에서 1.0(긍정)에 이르는 단일 이중만 제공positive했습니다. 단일 점수 대신 세 개의 이중 값을 허용하도록 인덱스 정의를 업데이트하고 모든 다운스트림 기술 입력, 지식 저장소 프로젝션 및 출력 필드 매핑이 명명 변경 내용과 일치하는지 확인해야 합니다.

이전 SentimentSkill을 SentimentSkill V3로 완전히 바꾸고, 다운스트림 기술 입력, 지식 저장소 프로젝션, 인덱서 출력 필드 매핑 및 인덱스 정의를 새 출력 형식과 일치하도록 업데이트하고, 인덱서를 다시 설정하여 모든 문서에 일관된 감정 결과가 전달되도록 하는 것이 좋습니다.

참고 항목

최신 버전의 감정 기술을 사용하기 위해 보강 파이프라인을 업데이트하는 데 추가적인 도움이 필요하거나 인덱서를 다시 설정하는 것이 옵션이 아닌 경우 직접 작업할 수 있는 새 지원 요청을세요.

Microsoft.Skills.Text.NamedEntityRecognitionSkill

마지막 사용 가능한 API 버전

2017-11-11-Preview

지원 종료

2024년 8월 31일 토요일

권장 사항

대신 Microsoft.Skills.Text.V3.EntityRecognitionSkill을 사용합니다. 이 기능은 NamedEntityRecognitionSkill의 대부분의 기능을 더 높은 품질로 제공합니다. 또한 복잡한 출력 필드에 다양한 정보가 있습니다.

Microsoft.Skills.Text.V3.EntityRecognitionSkill로 마이그레이션하려면 기술 정의를 다음 중 하나 이상 변경합니다. 기술 세트 업데이트 API사용하여 기술 정의를 업데이트할 수 있습니다.

  1. (필수)@odata.type"#Microsoft.Skills.Text.NamedEntityRecognitionSkill"에서 "#Microsoft.Skills.Text.V3.EntityRecognitionSkill"로 변경

  2. (선택 사항) 출력을 사용하는 entities 경우 대신 복합 컬렉션 출력을 EntityRecognitionSkill V3 사용합니다namedEntities. 새 namedEntities 복합 출력의 속성 이름에 몇 가지 사소한 변경 내용이 있습니다.

    1. value 이름이 로 바뀝니다. text
    2. confidence 이름이 로 바뀝니다. confidenceScore

    정확히 동일한 속성 이름을 생성해야 하는 경우 ShaperSkill추가하여 필요한 이름으로 출력의 모양을 바꿉니다. 예를 들어 이 ShaperSkill은 속성 이름을 이전 값으로 바꿉니다.

    {
        "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
        "name": "NamedEntitiesShaper",
        "description": "NamedEntitiesShaper",
        "context": "/document/namedEntities",
        "inputs": [
            {
                "name": "old_format",
                "sourceContext": "/document/namedEntities/*",
                "inputs": [
                    {
                        "name": "value",
                        "source": "/document/namedEntities/*/text"
                    },
                    {
                        "name": "offset",
                        "source": "/document/namedEntities/*/offset"
                    },
                    {
                        "name": "category",
                        "source": "/document/namedEntities/*/category"
                    },
                    {
                        "name": "confidence",
                        "source": "/document/namedEntities/*/confidenceScore"
                    }
                ]
            }
        ],
        "outputs": [
            {
                "name": "output",
                "targetName": "entities"
            }
        ]
    }
    
  3. (선택 사항) 명시적으로 지정 categoriesEntityRecognitionSkill V3 하지 않으면 에서 지원NamedEntityRecognitionSkill한 범주 외에 다른 유형의 범주를 반환할 수 있습니다. 이 동작이 바람직하지 않은 경우 명시적으로 categories 매개 변수를 ["Person", "Location", "Organization"]으로 설정해야 합니다.

    샘플 마이그레이션 정의

    • 간단한 마이그레이션

      (이전) NamedEntityRecognition 기술 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.NamedEntityRecognitionSkill",
          "categories": [ "Person" ],
          "defaultLanguageCode": "en",
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              }
          ]
      }
      

      (이후) EntityRecognition 기술 V3 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.V3.EntityRecognitionSkill",
          "categories": [ "Person" ],
          "defaultLanguageCode": "en",
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              }
          ]
      }
      
    • 약간 복잡한 마이그레이션

      (이전) NamedEntityRecognition 기술 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.NamedEntityRecognitionSkill",
          "defaultLanguageCode": "en",
          "minimumPrecision": 0.1,
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              },
              {
                  "name": "entities"
              }
          ]
      }
      

      (이후) EntityRecognition 기술 V3 정의

      {
          "@odata.type": "#Microsoft.Skills.Text.V3.EntityRecognitionSkill",
          "categories": [ "Person", "Location", "Organization" ],
          "defaultLanguageCode": "en",
          "minimumPrecision": 0.1,
          "inputs": [
              {
                  "name": "text",
                  "source": "/document/content"
              }
          ],
          "outputs": [
              {
                  "name": "persons",
                  "targetName": "people"
              },
              {
                  "name": "namedEntities"
              }
          ]
      },
      {
          "@odata.type": "#Microsoft.Skills.Util.ShaperSkill",
          "name": "NamedEntitiesShaper",
          "description": "NamedEntitiesShaper",
          "context": "/document/namedEntities",
          "inputs": [
              {
                  "name": "old_format",
                  "sourceContext": "/document/namedEntities/*",
                  "inputs": [
                      {
                          "name": "value",
                          "source": "/document/namedEntities/*/text"
                      },
                      {
                          "name": "offset",
                          "source": "/document/namedEntities/*/offset"
                      },
                      {
                          "name": "category",
                          "source": "/document/namedEntities/*/category"
                      },
                      {
                          "name": "confidence",
                          "source": "/document/namedEntities/*/confidenceScore"
                      }
                  ]
              }
          ],
          "outputs": [
              {
                  "name": "output",
                  "targetName": "entities"
              }
          ]
      }
      

참고 항목