azure-language-service

Azure AI Language 技能

此技能提供 Azure AI Language 方面的专家指导，涵盖故障排除、最佳实践、决策制定、架构与设计模式、限制与配额、安全性、配置、集成与编码模式，以及部署。它结合了本地快速参考内容与远程文档获取功能。

如何使用此技能

IMPORTANT for Agent：使用下方的类别索引查找相关部分。对于带有行号范围的类别（例如 L35-L120），使用 read_file 读取指定行。对于带有文件链接的类别（例如 [security.md](security.md)），使用 read_file 读取所链接的参考文件

IMPORTANT for Agent：如果 metadata.generated_at 的时间早于 3 个月以上，建议用户从仓库拉取最新版本。如果 mcp_microsoftdocs 工具不可用，建议用户安装：安装指南

此技能需要网络访问权限来获取文档内容：

• 首选：使用 mcp_microsoftdocs:microsoft_docs_fetch，并传入查询字符串 from=learn-agent-skill。返回 Markdown。
• 备用：使用 fetch_webpage，并传入查询字符串 from=learn-agent-skill&accept=text/markdown。返回 Markdown。

类别索引

类别	行号	描述
故障排除	L37-L42	诊断和解决 Azure Language 自定义 NER 和对话式问答 (CQA) 中的常见问题，包括模型错误、配置问题和故障排除工作流。
最佳实践	L43-L53	设计和编写 CLU、自定义 NER、PII 和 CQA 项目的最佳实践，包括数据准备、架构、生命周期、闲聊角色设定和文档格式。
决策制定	L54-L63	关于选择区域和应用类型、规划 CQA 解决方案，以及决定或执行从 LUIS、QnA Maker、Text Analytics 和 Language Studio 迁移到 Azure Language/Fountry 的指南。
架构与设计模式	L64-L71	设计和实现 CLU、自定义 NER、自定义文本分类及 Azure AI Language 中编排工作流模型的区域故障转移和高可用性模式。
限制与配额	L72-L95	Azure Language 功能（CLU、NER、文本分类、CQA、健康）的限制、配额、语言和模型生命周期，包括数据大小、速率、吞吐量和容器限制。
安全性	L96-L107	保护 Azure Language 和 CQA：静态加密（包括 CMK）、RBAC、托管标识、SAS 令牌、网络隔离/Private Link，以及安全部署和数据访问配置。
配置	L108-L124	配置 Azure AI Language 功能：CLU 微调、容器、自定义/健康 NER、编排 None 意图、CQA 评分/遥测、FHIR 输出，以及相关项目/技能设置。
集成与编码模式	L125-L144	如何调用 Azure Language/TA/health/CLU/CQA API 和 SDK，以实现 NER、实体链接、关键短语、语言检测、情感分析、PII 编辑、关系和编排工作流。
部署	L145-L155	关于跨区域部署和运行自定义语言/NER/CQA/情感分析/健康模型、通过 Docker 容器在本地部署，以及在环境之间移动项目的指南。

故障排除

主题	URL
解决 Azure Language 自定义 NER 的常见问题	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/faq
排查常见 CQA 问题和错误	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/troubleshooting

最佳实践

主题	URL
遵循 CLU 项目编写最佳实践	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/concepts/best-practices
让 PII 检测适应自定义领域术语	https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/how-to/adapt-to-domain-pii
遵循 CQA 项目最佳实践	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/best-practices
遵循 CQA 项目开发生命周期	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/project-development-lifecycle
在 CQA 项目中应用编写最佳实践	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/best-practices
在 CQA 项目中添加闲聊角色	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/chit-chat
为自定义问答应用文档格式设置最佳实践	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/reference/document-format-guidelines

决策

主题	URL
为 Language 服务功能选择 Azure 区域	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/regional-support
选择 CLU 应用或编排工作流	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/concepts/app-architecture
将 Azure Language Studio 项目迁移到 Microsoft Foundry	https://learn.microsoft.com/en-us/azure/ai-services/language-service/migration-studio-to-foundry
规划 CQA 应用并选择 Azure 资源	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/plan
决定是否将 LUIS 和 QnA Maker 迁移到 Azure Language	https://learn.microsoft.com/en-us/azure/ai-services/language-service/reference/migrate
将 Text Analytics 应用迁移到 Azure Language API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/reference/migrate-language-service-latest

架构与设计模式

主题	URL
为 CLU 模型设计区域故障转移	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/how-to/fail-over
为自定义 NER 模型设计区域故障转移	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/fail-over
为自定义文本分类设计区域故障转移	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-text-classification/fail-over
为编排工作流模型实现区域故障转移	https://learn.microsoft.com/en-us/azure/ai-services/language-service/orchestration-workflow/concepts/fail-over

限制与配额

主题	URL
Azure Language 功能的数据大小和速率限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/data-limits
了解 Azure Language 模型的生命周期时间线	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/model-lifecycle
使用 CLU 容器及其本地部署限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/how-to/use-containers
查看 CLU 支持的语言和区域设置	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/language-support
查阅 CLU 预生成实体组件参考	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/prebuilt-component-reference
查看 CLU 数据量和吞吐量限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/service-limits
管理自定义 NER 训练作业及其过期时间	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/how-to/train-model
查看自定义文本分类的语言支持矩阵	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-text-classification/language-support
查看自定义文本分类的数据量和速率限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-text-classification/service-limits
查看实体链接的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/entity-linking/language-support
查看关键短语提取的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/key-phrase-extraction/language-support
了解 NER 中的实体类别和类型	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/concepts/named-entity-categories
查看 Azure NER 功能的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/language-support
查看业务流程工作流项目的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/orchestration-workflow/language-support
查看业务流程工作流的数据量和吞吐量限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/orchestration-workflow/service-limits
了解 PII 检测容器的数据限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/how-to/use-containers
了解 CQA 项目和服务限制	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/limits
查看 CQA 项目的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/language-support
探索健康分析中的医疗实体类别	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/concepts/health-entity-categories
查看 Text Analytics for health 的语言支持	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/language-support

安全性

主题	URL
了解 Azure Language 中静态数据加密	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/encryption-data-at-rest
为 Azure Language 资源配置 RBAC	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/role-based-access-control
使用托管标识保护存储 Blob 访问	https://learn.microsoft.com/en-us/azure/ai-services/language-service/native-document-support/managed-identities
为 Azure 存储 Blob 配置 SAS 令牌	https://learn.microsoft.com/en-us/azure/ai-services/language-service/native-document-support/shared-access-signatures
为 CQA 微调配置 Azure 资源	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/configure-azure-resources
为静态 CQA 数据配置 CMK 加密	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/encrypt-data-at-rest
为 CQA 启用网络隔离和 Private Link	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/network-isolation
保护 Azure Language 部署和数据访问	https://learn.microsoft.com/en-us/azure/ai-services/language-service/secure-deployment

配置

主题	URL
为 CLU 微调模型配置 Azure 资源	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/configure-azure-resources
配置 Azure AI Language 服务容器	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/configure-containers
使用自定义 NER 支持的数据格式	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/concepts/data-formats
使用 NER 实体元数据和解析结果	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/concepts/entity-metadata
映射 NER API 版本和实体标签／类型	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/concepts/ga-preview-mapping
配置 NER 技能参数和推理选项	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/how-to/skill-parameters
在编排工作流中配置 None 意图行为	https://learn.microsoft.com/en-us/azure/ai-services/language-service/orchestration-workflow/concepts/none-intent
解读和配置 CQA 置信度分数	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/confidence-score
启用并查询 CQA 分析遥测数据	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/analytics
管理 CQA 项目设置和源	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/manage-knowledge-base
在 Text Analytics for health 中使用断言检测	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/concepts/assertion-detection
在健康分析输出中配置 FHIR 结构化处理	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/concepts/fhir
配置 Text Analytics for health 容器设置	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/how-to/configure-containers

集成与编码模式

主题	URL
通过 SDK 和 REST API 集成 Azure Language	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/developer-guide
调用自定义 NER 预测 API 和 SDK	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/how-to/call-api
通过 Foundry 或 REST 开始构建自定义 NER 模型	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/quickstart
使用正确的参数调用实体链接 API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/entity-linking/how-to/call-api
正确调用 Key Phrase Extraction API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/key-phrase-extraction/how-to/call-api
使用 SDK 和 REST 实现语言检测	https://learn.microsoft.com/en-us/azure/ai-services/language-service/language-detection/quickstart
使用 Azure Language 服务执行 NER 调用	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/how-to-call
使用 NER 客户端库构建 .NET 应用	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/quickstart
通过编排工作流集成 CLU 和自定义 Q&A	https://learn.microsoft.com/en-us/azure/ai-services/language-service/orchestration-workflow/tutorials/connect-services
使用 Azure Language 从原生文档中删除 PII	https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/how-to/redact-document-pii
使用 REST API 自动化 CQA 编写	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/authoring
使用 CQA 预构建的问答 API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/prebuilt
正确调用情感分析和意见挖掘 API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/sentiment-opinion-mining/how-to/call-api
在 Text Analytics for health 中使用关系提取	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/concepts/relation-extraction
调用 Text Analytics for health API	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/how-to/call-api
集成 Text Analytics for health 的快速入门	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/quickstart

部署

主题	URL
将自定义语言项目部署到多个区域	https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/custom-features/multi-region-deployment
使用 Docker 容器在本地部署自定义 NER	https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-named-entity-recognition/how-to/use-containers
在本地部署 Key Phrase Extraction 容器	https://learn.microsoft.com/en-us/azure/ai-services/language-service/key-phrase-extraction/how-to/use-containers
在本地运行 NER Docker 容器	https://learn.microsoft.com/en-us/azure/ai-services/language-service/named-entity-recognition/how-to/use-containers
在 Foundry 中创建并部署 CQA Agent	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/deploy-agent
在不同环境之间迁移 CQA 项目	https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/migrate-knowledge-base
在本地部署 Sentiment Analysis 容器	https://learn.microsoft.com/en-us/azure/ai-services/language-service/sentiment-opinion-mining/how-to/use-containers
在本地部署 Text Analytics for health 容器	https://learn.microsoft.com/en-us/azure/ai-services/language-service/text-analytics-for-health/how-to/use-containers