DataMate/frontend/src/pages/DataAnnotation/annotation.model.ts

import type { DatasetType } from "@/pages/DataManagement/dataset.model";

export enum AnnotationTaskStatus {
  ACTIVE = "active",
  INACTIVE = "inactive",
  PROCESSING = "processing",
  COMPLETED = "completed",
  SKIPPED = "skipped",
}

export enum AnnotationResultStatus {
  ANNOTATED = "ANNOTATED",
  IN_PROGRESS = "IN_PROGRESS",
  NO_ANNOTATION = "NO_ANNOTATION",
  NOT_APPLICABLE = "NOT_APPLICABLE",
}

export interface AnnotationTask {
  id: string;
  name: string;
  labelingProjId: string;
  datasetId: string;

  annotationCount: number;

  description?: string;
  assignedTo?: string;
  progress: number;
  statistics: {
    accuracy: number;
    averageTime: number;
    reviewCount: number;
  };
  status: AnnotationTaskStatus;
  totalDataCount: number;
  type: DatasetType;

  createdAt: string;
  updatedAt: string;
}

// 标注模板相关类型
export interface LabelDefinition {
  fromName: string;
  toName: string;
  type: string;
  options?: string[];
  labels?: string[];
  required?: boolean;
  description?: string;
}

export interface ObjectDefinition {
  name: string;
  type: string;
  value: string;
}

export interface TemplateConfiguration {
  labels: LabelDefinition[];
  objects: ObjectDefinition[];
  metadata?: Record<string, unknown>;
}

export interface AnnotationTemplate {
  id: string;
  name: string;
  description?: string;
  dataType: string;
  labelingType: string;
  configuration?: TemplateConfiguration;
  labelConfig?: string;
  style: string;
  category: string;
  builtIn: boolean;
  version: string;
  createdAt: string;
  updatedAt?: string;
}

export interface AnnotationTemplateListResponse {
  content: AnnotationTemplate[];
  total: number;
  page: number;
  size: number;
  totalPages: number;
}

/**
 * 数据类型枚举
 * 定义标注任务支持的数据类型
 */
export enum DataType {
  TEXT = "text",
  IMAGE = "image",
  AUDIO = "audio",
  VIDEO = "video",
  PDF = "pdf",
  TIMESERIES = "timeseries",
  CHAT = "chat",
  HTML = "html",
  TABLE = "table",
}

/**
 * 模板分类枚举
 * 基于 Label Studio 模板分类体系
 */
export enum Classification {
  // 音频/语音处理
  AUDIO_SPEECH = "audio-speech",
  // 聊天评估
  CHAT = "chat",
  // 计算机视觉
  COMPUTER_VISION = "computer-vision",
  // 对话AI
  CONVERSATIONAL_AI = "conversational-ai",
  // 生成式AI
  GENERATIVE_AI = "generative-ai",
  // 自然语言处理
  NLP = "nlp",
  // 排名与评分
  RANKING_SCORING = "ranking-scoring",
  // 结构化数据解析
  STRUCTURED_DATA = "structured-data",
  // 时间序列分析
  TIME_SERIES = "time-series",
  // 视频处理
  VIDEO = "video",
  // 社区贡献
  COMMUNITY = "community",
  // 自定义
  CUSTOM = "custom",
}

/**
 * 标注类型枚举
 * 定义各种具体的标注任务类型
 */
export enum AnnotationType {
  // ===== 音频/语音处理 =====
  // 自动语音识别（分段）
  ASR_SEGMENTS = "asr-segments",
  // 自动语音识别
  ASR = "asr",
  // 对话分析
  CONVERSATION_ANALYSIS = "conversation-analysis",
  // 意图分类
  INTENT_CLASSIFICATION = "intent-classification",
  // 信号质量检测
  SIGNAL_QUALITY = "signal-quality",
  // 声音事件检测
  SOUND_EVENT_DETECTION = "sound-event-detection",
  // 说话人分割
  SPEAKER_SEGMENTATION = "speaker-segmentation",
  // 语音转录
  SPEECH_TRANSCRIPTION = "speech-transcription",

  // ===== 聊天评估 =====
  // 无LLM的Agent微调
  AGENT_FINE_TUNE = "agent-fine-tune",
  // 有LLM的Agent微调
  AGENT_FINE_TUNE_LLM = "agent-fine-tune-llm",
  // 红队测试
  RED_TEAMING = "red-teaming",
  // RLHF生产对话评估
  RLHF_EVALUATION = "rlhf-evaluation",
  // 聊天机器人评估
  CHATBOT_EVALUATION = "chatbot-evaluation",

  // ===== 计算机视觉 =====
  // 图像描述
  IMAGE_CAPTIONING = "image-captioning",
  // 图像分类
  IMAGE_CLASSIFICATION = "image-classification",
  // 库存追踪
  INVENTORY_TRACKING = "inventory-tracking",
  // 关键点标注
  KEYPOINT_LABELING = "keypoint-labeling",
  // 医学图像分类
  MEDICAL_IMAGE_CLASSIFICATION = "medical-image-classification",
  // 多页文档标注
  MULTIPAGE_DOCUMENT = "multipage-document",
  // 目标检测（边界框）
  OBJECT_DETECTION = "object-detection",
  // OCR识别
  OCR = "ocr",
  // PDF OCR标注
  PDF_OCR = "pdf-ocr",
  // 语义分割（掩码）
  SEMANTIC_SEGMENTATION_MASK = "semantic-segmentation-mask",
  // 语义分割（多边形）
  SEMANTIC_SEGMENTATION_POLYGON = "semantic-segmentation-polygon",
  // Visual Genome
  VISUAL_GENOME = "visual-genome",
  // 视觉问答
  VQA = "vqa",

  // ===== 对话AI =====
  // 共指消解与实体链接
  COREFERENCE_RESOLUTION = "coreference-resolution",
  // 意图分类与槽填充
  SLOT_FILLING = "slot-filling",
  // 响应生成
  RESPONSE_GENERATION = "response-generation",
  // 响应选择
  RESPONSE_SELECTION = "response-selection",

  // ===== 生成式AI =====
  // 聊天机器人模型评估
  CHATBOT_ASSESSMENT = "chatbot-assessment",
  // RLHF人类偏好收集
  RLHF_PREFERENCE = "rlhf-preference",
  // LLM排名
  LLM_RANKING = "llm-ranking",
  // LLM响应评分
  LLM_GRADING = "llm-grading",
  // 监督微调
  SFT = "sft",
  // 视觉排名
  VISUAL_RANKING = "visual-ranking",

  // ===== 自然语言处理 =====
  // 内容审核
  CONTENT_MODERATION = "content-moderation",
  // 机器翻译
  MACHINE_TRANSLATION = "machine-translation",
  // 命名实体识别
  NER = "ner",
  // 问答
  QUESTION_ANSWERING = "question-answering",
  // 关系抽取
  RELATION_EXTRACTION = "relation-extraction",
  // 分类法/层级分类
  TAXONOMY = "taxonomy",
  // 文本分类
  TEXT_CLASSIFICATION = "text-classification",
  // 文本摘要
  TEXT_SUMMARIZATION = "text-summarization",

  // ===== 排名与评分 =====
  // ASR假设选择
  ASR_HYPOTHESES = "asr-hypotheses",
  // 基于内容的图像检索
  IMAGE_RETRIEVAL = "image-retrieval",
  // 文档检索
  DOCUMENT_RETRIEVAL = "document-retrieval",
  // 成对分类
  PAIRWISE_CLASSIFICATION = "pairwise-classification",
  // 成对回归
  PAIRWISE_REGRESSION = "pairwise-regression",
  // 搜索页面排名
  SERP_RANKING = "serp-ranking",
  // 文本到图像生成
  TEXT_TO_IMAGE = "text-to-image",

  // ===== 结构化数据解析 =====
  // 自由格式元数据
  FREEFORM_METADATA = "freeform-metadata",
  // HTML实体识别
  HTML_ENTITY_RECOGNITION = "html-entity-recognition",
  // PDF分类
  PDF_CLASSIFICATION = "pdf-classification",
  // 表格数据标注
  TABULAR_DATA = "tabular-data",

  // ===== 时间序列分析 =====
  // 活动识别
  ACTIVITY_RECOGNITION = "activity-recognition",
  // 变点检测
  CHANGE_POINT_DETECTION = "change-point-detection",
  // 异常检测
  ANOMALY_DETECTION = "anomaly-detection",
  // 时间序列信号质量
  TIMESERIES_SIGNAL_QUALITY = "timeseries-signal-quality",
  // 时间序列预测
  TIMESERIES_FORECASTING = "timeseries-forecasting",

  // ===== 视频处理 =====
  // 视频分类
  VIDEO_CLASSIFICATION = "video-classification",
  // 视频帧分类
  VIDEO_FRAME_CLASSIFICATION = "video-frame-classification",
  // 视频目标追踪
  VIDEO_OBJECT_TRACKING = "video-object-tracking",
  // 视频时间线分割
  VIDEO_TIMELINE_SEGMENTATION = "video-timeline-segmentation",

  // ===== 社区贡献 =====
  // 乳腺癌乳房X光分类
  MAMMOGRAM_CLASSIFICATION = "mammogram-classification",
  // HTML NER标注
  HTML_NER_TAGGING = "html-ner-tagging",
  // 发票NER标注（BIO格式）
  INVOICE_NER_BIO = "invoice-ner-bio",
  // OCR发票预NER（BIO格式）
  OCR_INVOICE_PRE_NER = "ocr-invoice-pre-ner",
  // Twitter/X情感分析
  TWITTER_SENTIMENT = "twitter-sentiment",

  // ===== 通用/遗留类型 =====
  // 通用分类
  CLASSIFICATION = "classification",
  // 通用分割
  SEGMENTATION = "segmentation",
}

export enum TemplateType {
  SYSTEM = "true",
  CUSTOM = "false"
}