@nice2dev/ui-tools 1.0.10

package/dist/dataset-management.d.ts

@@ -0,0 +1,782 @@
+/**
+ * @file dataset-management.ts
+ * @module @nice2dev/ui-tools
+ * @description PRO-18.1 — Dataset Management
+ *
+ * Types for ML/AI dataset management: import, labeling, versioning,
+ * annotation, quality control, data augmentation, active learning.
+ */
+/** Dataset type */
+export type DatasetType = 'tabular' | 'image' | 'text' | 'audio' | 'video' | 'timeSeries' | 'pointCloud' | 'multimodal' | 'custom';
+/** Data format */
+export type DataFormat = 'csv' | 'json' | 'jsonl' | 'parquet' | 'arrow' | 'avro' | 'tfrecord' | 'hdf5' | 'pickle' | 'images' | 'audio' | 'video' | 'custom';
+/** Label type */
+export type LabelType = 'classification' | 'multiLabel' | 'boundingBox' | 'polygon' | 'segmentation' | 'keypoint' | 'ner' | 'sentiment' | 'transcription' | 'translation' | 'qa' | 'textSpan' | 'segment' | 'speaker' | 'custom';
+/** Data split */
+export type DataSplit = 'train' | 'validation' | 'test' | 'holdout';
+/** Annotation status */
+export type AnnotationStatus = 'pending' | 'inProgress' | 'completed' | 'review' | 'approved' | 'rejected';
+/** Dataset browser configuration */
+export interface DatasetBrowserConfig {
+    /** Theme */
+    theme?: DatasetTheme;
+    /** Features */
+    features?: DatasetFeatures;
+    /** Storage */
+    storage?: DatasetStorage;
+    /** Labeling config */
+    labeling?: LabelingConfig;
+    /** Versioning config */
+    versioning?: VersioningConfig;
+}
+/** Dataset theme */
+export interface DatasetTheme {
+    name?: string;
+    colors?: DatasetColors;
+    fonts?: DatasetFonts;
+}
+/** Dataset colors */
+export interface DatasetColors {
+    primary?: string;
+    secondary?: string;
+    background?: string;
+    surface?: string;
+    labelColors?: string[];
+}
+/** Dataset fonts */
+export interface DatasetFonts {
+    family?: string;
+    monoFamily?: string;
+}
+/** Dataset features */
+export interface DatasetFeatures {
+    import?: boolean;
+    export?: boolean;
+    labeling?: boolean;
+    versioning?: boolean;
+    augmentation?: boolean;
+    activeLearning?: boolean;
+    statistics?: boolean;
+    preview?: boolean;
+    search?: boolean;
+    filter?: boolean;
+}
+/** Dataset storage */
+export interface DatasetStorage {
+    type: 's3' | 'gcs' | 'azure' | 'local' | 'custom';
+    bucket?: string;
+    prefix?: string;
+    credentials?: StorageCredentials;
+    endpoint?: string;
+}
+/** Storage credentials */
+export interface StorageCredentials {
+    accessKeyId?: string;
+    secretAccessKey?: string;
+    region?: string;
+    accountName?: string;
+    accountKey?: string;
+    connectionString?: string;
+}
+/** Labeling config */
+export interface LabelingConfig {
+    enabled?: boolean;
+    labelSchema?: LabelSchema;
+    reviewRequired?: boolean;
+    qualityThreshold?: number;
+    activeLearn?: ActiveLearningConfig;
+}
+/** Label schema */
+export interface LabelSchema {
+    type: LabelType;
+    classes?: LabelClass[];
+    attributes?: LabelAttribute[];
+    instructions?: string;
+    examples?: LabelExample[];
+}
+/** Label class */
+export interface LabelClass {
+    id: string;
+    name: string;
+    color?: string;
+    description?: string;
+    shortcut?: string;
+    parent?: string;
+}
+/** Label attribute */
+export interface LabelAttribute {
+    id: string;
+    name: string;
+    type: 'text' | 'number' | 'boolean' | 'select' | 'multiSelect';
+    required?: boolean;
+    options?: string[];
+    default?: unknown;
+}
+/** Label example */
+export interface LabelExample {
+    id: string;
+    data: unknown;
+    labels: Label[];
+    explanation?: string;
+}
+/** Active learning config */
+export interface ActiveLearningConfig {
+    enabled?: boolean;
+    strategy: ActiveLearningStrategy;
+    batchSize?: number;
+    threshold?: number;
+    modelId?: string;
+}
+/** Active learning strategy */
+export type ActiveLearningStrategy = 'uncertainty' | 'margin' | 'entropy' | 'diversity' | 'representativeness' | 'custom';
+/** Versioning config */
+export interface VersioningConfig {
+    enabled?: boolean;
+    backend: 'dvc' | 'pachyderm' | 'lakefs' | 'delta' | 'custom';
+    remote?: string;
+    autoVersion?: boolean;
+}
+/** Dataset */
+export interface Dataset {
+    id: string;
+    name: string;
+    description?: string;
+    type: DatasetType;
+    format: DataFormat;
+    schema?: DatasetSchema;
+    statistics?: DatasetStatistics;
+    splits?: DatasetSplitInfo[];
+    version?: DatasetVersion;
+    labels?: LabelSchema;
+    tags?: string[];
+    metadata?: DatasetMetadata;
+}
+/** Dataset schema */
+export interface DatasetSchema {
+    columns?: ColumnSchema[];
+    targetColumn?: string;
+    featureColumns?: string[];
+    mediaFields?: MediaFieldSchema[];
+}
+/** Column schema */
+export interface ColumnSchema {
+    name: string;
+    type: ColumnDataType;
+    nullable?: boolean;
+    description?: string;
+    statistics?: ColumnStatistics;
+}
+/** Column data type */
+export type ColumnDataType = 'string' | 'integer' | 'float' | 'boolean' | 'datetime' | 'date' | 'array' | 'object' | 'binary';
+/** Column statistics */
+export interface ColumnStatistics {
+    count?: number;
+    nullCount?: number;
+    uniqueCount?: number;
+    min?: number | string;
+    max?: number | string;
+    mean?: number;
+    std?: number;
+    median?: number;
+    mode?: unknown;
+    histogram?: HistogramBin[];
+    topValues?: ValueCount[];
+}
+/** Histogram bin */
+export interface HistogramBin {
+    low: number;
+    high: number;
+    count: number;
+}
+/** Value count */
+export interface ValueCount {
+    value: unknown;
+    count: number;
+    percentage?: number;
+}
+/** Media field schema */
+export interface MediaFieldSchema {
+    name: string;
+    type: 'image' | 'audio' | 'video' | 'document';
+    format?: string[];
+    maxSize?: number;
+}
+/** Dataset statistics */
+export interface DatasetStatistics {
+    totalSamples: number;
+    labeledSamples?: number;
+    unlabeledSamples?: number;
+    totalSize?: number;
+    averageSampleSize?: number;
+    classDistribution?: ClassDistribution[];
+    labelQuality?: LabelQualityMetrics;
+    createdAt?: Date;
+    updatedAt?: Date;
+}
+/** Class distribution */
+export interface ClassDistribution {
+    classId: string;
+    className?: string;
+    count: number;
+    percentage: number;
+    split?: DataSplit;
+}
+/** Label quality metrics */
+export interface LabelQualityMetrics {
+    agreementRate?: number;
+    interAnnotatorKappa?: number;
+    averageConfidence?: number;
+    reviewedPercentage?: number;
+    rejectionRate?: number;
+}
+/** Dataset split info */
+export interface DatasetSplitInfo {
+    split: DataSplit;
+    count: number;
+    percentage: number;
+    path?: string;
+}
+/** Dataset version */
+export interface DatasetVersion {
+    version: string;
+    commit?: string;
+    timestamp: Date;
+    author?: string;
+    message?: string;
+    parent?: string;
+    size?: number;
+    checksum?: string;
+}
+/** Dataset metadata */
+export interface DatasetMetadata {
+    source?: string;
+    license?: string;
+    citation?: string;
+    domain?: string;
+    language?: string[];
+    createdBy?: string;
+    createdAt?: Date;
+    modifiedBy?: string;
+    modifiedAt?: Date;
+    custom?: Record<string, unknown>;
+}
+/** Import task */
+export interface ImportTask {
+    id: string;
+    datasetId: string;
+    source: ImportSource;
+    options: ImportOptions;
+    status: ImportStatus;
+    progress?: ImportProgress;
+    result?: ImportResult;
+    error?: string;
+    startedAt?: Date;
+    completedAt?: Date;
+}
+/** Import source */
+export interface ImportSource {
+    type: ImportSourceType;
+    path?: string;
+    url?: string;
+    bucket?: string;
+    prefix?: string;
+    connection?: string;
+    query?: string;
+}
+/** Import source type */
+export type ImportSourceType = 'local' | 'url' | 's3' | 'gcs' | 'azure' | 'huggingface' | 'kaggle' | 'database' | 'api';
+/** Import options */
+export interface ImportOptions {
+    format?: DataFormat;
+    encoding?: string;
+    delimiter?: string;
+    headers?: boolean;
+    skipRows?: number;
+    limitRows?: number;
+    columns?: string[];
+    schema?: DatasetSchema;
+    dedup?: boolean;
+    validate?: boolean;
+    transform?: DataTransform[];
+}
+/** Data transform */
+export interface DataTransform {
+    type: TransformType;
+    column?: string;
+    config: TransformConfig;
+}
+/** Transform type */
+export type TransformType = 'rename' | 'drop' | 'cast' | 'fillna' | 'normalize' | 'encode' | 'split' | 'merge' | 'filter' | 'custom';
+/** Transform config */
+export interface TransformConfig {
+    newName?: string;
+    targetType?: ColumnDataType;
+    fillValue?: unknown;
+    method?: string;
+    pattern?: string;
+    expression?: string;
+    [key: string]: unknown;
+}
+/** Import status */
+export type ImportStatus = 'pending' | 'validating' | 'importing' | 'transforming' | 'completed' | 'failed' | 'cancelled';
+/** Import progress */
+export interface ImportProgress {
+    stage: string;
+    current: number;
+    total: number;
+    percentage: number;
+    bytesProcessed?: number;
+    rowsProcessed?: number;
+    errors?: number;
+}
+/** Import result */
+export interface ImportResult {
+    rowsImported: number;
+    rowsSkipped: number;
+    rowsFailed: number;
+    warnings?: ImportWarning[];
+    schema?: DatasetSchema;
+}
+/** Import warning */
+export interface ImportWarning {
+    row?: number;
+    column?: string;
+    message: string;
+    severity: 'info' | 'warning' | 'error';
+}
+/** Labeling task */
+export interface LabelingTask {
+    id: string;
+    datasetId: string;
+    name: string;
+    description?: string;
+    schema: LabelSchema;
+    samples: LabelingSample[];
+    assignees?: LabelingAssignee[];
+    progress?: LabelingProgress;
+    config?: LabelingTaskConfig;
+    createdAt: Date;
+    deadline?: Date;
+}
+/** Labeling sample */
+export interface LabelingSample {
+    id: string;
+    data: unknown;
+    labels?: Label[];
+    status: AnnotationStatus;
+    assignee?: string;
+    confidence?: number;
+    review?: LabelReview;
+    history?: LabelHistory[];
+}
+/** Label */
+export interface Label {
+    id: string;
+    type: LabelType;
+    classId?: string;
+    className?: string;
+    value?: unknown;
+    attributes?: Record<string, unknown>;
+    geometry?: LabelGeometry;
+    spans?: TextSpan[];
+    confidence?: number;
+    createdBy?: string;
+    createdAt?: Date;
+}
+/** Label geometry */
+export interface LabelGeometry {
+    type: GeometryType;
+    coordinates: number[];
+    points?: Point2D[];
+    bbox?: BoundingBox;
+    mask?: SegmentationMask;
+    keypoints?: Keypoint[];
+}
+/** Geometry type */
+export type GeometryType = 'point' | 'rectangle' | 'polygon' | 'polyline' | 'ellipse' | 'mask' | 'keypoint';
+/** Point 2D */
+export interface Point2D {
+    x: number;
+    y: number;
+}
+/** Bounding box */
+export interface BoundingBox {
+    x: number;
+    y: number;
+    width: number;
+    height: number;
+    rotation?: number;
+}
+/** Segmentation mask */
+export interface SegmentationMask {
+    format: 'rle' | 'polygon' | 'bitmap';
+    data: string | number[][];
+    width?: number;
+    height?: number;
+}
+/** Keypoint */
+export interface Keypoint {
+    id: string;
+    name?: string;
+    x: number;
+    y: number;
+    visible?: boolean;
+    confidence?: number;
+}
+/** Text span */
+export interface TextSpan {
+    start: number;
+    end: number;
+    text?: string;
+    labels: string[];
+}
+/** Label review */
+export interface LabelReview {
+    status: 'approved' | 'rejected' | 'changes_requested';
+    reviewer: string;
+    timestamp: Date;
+    comments?: string;
+    corrections?: Label[];
+}
+/** Label history */
+export interface LabelHistory {
+    labels: Label[];
+    author: string;
+    timestamp: Date;
+    action: 'create' | 'update' | 'delete';
+}
+/** Labeling assignee */
+export interface LabelingAssignee {
+    userId: string;
+    name?: string;
+    role: 'annotator' | 'reviewer' | 'admin';
+    assigned: number;
+    completed: number;
+    quality?: number;
+}
+/** Labeling progress */
+export interface LabelingProgress {
+    total: number;
+    pending: number;
+    inProgress: number;
+    completed: number;
+    review: number;
+    approved: number;
+    rejected: number;
+    percentage: number;
+}
+/** Labeling task config */
+export interface LabelingTaskConfig {
+    allowOverlap?: boolean;
+    requireReview?: boolean;
+    minAnnotations?: number;
+    consensusThreshold?: number;
+    showPredictions?: boolean;
+    assistedLabeling?: boolean;
+    shortcuts?: LabelShortcut[];
+}
+/** Label shortcut */
+export interface LabelShortcut {
+    key: string;
+    action: string;
+    classId?: string;
+}
+/** Image annotation */
+export interface ImageAnnotation {
+    id: string;
+    imageId: string;
+    imageUrl: string;
+    width: number;
+    height: number;
+    labels: ImageLabel[];
+    status: AnnotationStatus;
+    metadata?: ImageMetadata;
+}
+/** Image label */
+export interface ImageLabel extends Label {
+    type: 'boundingBox' | 'polygon' | 'segmentation' | 'keypoint';
+    geometry: LabelGeometry;
+    occluded?: boolean;
+    truncated?: boolean;
+    difficult?: boolean;
+}
+/** Image metadata */
+export interface ImageMetadata {
+    filename?: string;
+    format?: string;
+    size?: number;
+    width: number;
+    height: number;
+    channels?: number;
+    exif?: Record<string, unknown>;
+}
+/** Text annotation */
+export interface TextAnnotation {
+    id: string;
+    textId: string;
+    text: string;
+    labels: TextLabel[];
+    status: AnnotationStatus;
+    metadata?: TextMetadata;
+}
+/** Text label */
+export interface TextLabel extends Label {
+    type: 'ner' | 'sentiment' | 'classification' | 'textSpan' | 'qa';
+    span?: TextSpan;
+    sentiment?: SentimentLabel;
+    qaAnswer?: QAAnswer;
+}
+/** Sentiment label */
+export interface SentimentLabel {
+    polarity: 'positive' | 'negative' | 'neutral' | 'mixed';
+    score?: number;
+    aspect?: string;
+}
+/** QA answer */
+export interface QAAnswer {
+    question: string;
+    answer: string;
+    start?: number;
+    end?: number;
+    answerType?: 'extractive' | 'abstractive' | 'yes_no' | 'unanswerable';
+}
+/** Text metadata */
+export interface TextMetadata {
+    source?: string;
+    language?: string;
+    tokenCount?: number;
+    sentenceCount?: number;
+}
+/** Audio annotation */
+export interface AudioAnnotation {
+    id: string;
+    audioId: string;
+    audioUrl: string;
+    duration: number;
+    sampleRate?: number;
+    labels: AudioLabel[];
+    status: AnnotationStatus;
+    metadata?: AudioMetadata;
+}
+/** Audio label */
+export interface AudioLabel extends Label {
+    type: 'transcription' | 'classification' | 'segment' | 'speaker';
+    segment?: AudioSegment;
+    transcription?: TranscriptionLabel;
+    speaker?: SpeakerLabel;
+}
+/** Audio segment */
+export interface AudioSegment {
+    start: number;
+    end: number;
+    channel?: number;
+}
+/** Transcription label */
+export interface TranscriptionLabel {
+    text: string;
+    words?: TranscriptionWord[];
+    language?: string;
+    confidence?: number;
+}
+/** Transcription word */
+export interface TranscriptionWord {
+    word: string;
+    start: number;
+    end: number;
+    confidence?: number;
+}
+/** Speaker label */
+export interface SpeakerLabel {
+    speakerId: string;
+    speakerName?: string;
+}
+/** Audio metadata */
+export interface AudioMetadata {
+    filename?: string;
+    format?: string;
+    size?: number;
+    duration: number;
+    sampleRate?: number;
+    channels?: number;
+    bitrate?: number;
+}
+/** Augmentation pipeline */
+export interface AugmentationPipeline {
+    id: string;
+    name: string;
+    description?: string;
+    dataType: DatasetType;
+    transforms: AugmentationTransform[];
+    probability?: number;
+    preview?: AugmentationPreview;
+}
+/** Augmentation transform */
+export interface AugmentationTransform {
+    id: string;
+    type: AugmentationType;
+    params: AugmentationParams;
+    probability?: number;
+    enabled?: boolean;
+}
+/** Augmentation type */
+export type AugmentationType = 'rotate' | 'flip' | 'crop' | 'resize' | 'brightness' | 'contrast' | 'saturation' | 'hue' | 'blur' | 'noise' | 'cutout' | 'mixup' | 'cutmix' | 'synonym' | 'backtranslation' | 'insertion' | 'deletion' | 'swap' | 'timeStretch' | 'pitchShift' | 'addNoise' | 'timeShift' | 'smote' | 'adasyn' | 'perturbation';
+/** Augmentation params */
+export interface AugmentationParams {
+    angle?: number;
+    scale?: number;
+    factor?: number;
+    kernel?: number;
+    sigma?: number;
+    size?: number;
+    ratio?: number;
+    alpha?: number;
+    language?: string;
+    numWords?: number;
+    rate?: number;
+    semitones?: number;
+    snr?: number;
+    shift?: number;
+    min?: number;
+    max?: number;
+    [key: string]: unknown;
+}
+/** Augmentation preview */
+export interface AugmentationPreview {
+    original: unknown;
+    augmented: unknown[];
+    params?: Record<string, unknown>;
+}
+/** Dataset version info */
+export interface DatasetVersionInfo {
+    datasetId: string;
+    versions: DatasetVersion[];
+    current: string;
+    branches?: DatasetBranch[];
+    tags?: DatasetTag[];
+}
+/** Dataset branch */
+export interface DatasetBranch {
+    name: string;
+    head: string;
+    protected?: boolean;
+    createdAt?: Date;
+}
+/** Dataset tag */
+export interface DatasetTag {
+    name: string;
+    version: string;
+    message?: string;
+    createdAt?: Date;
+    createdBy?: string;
+}
+/** Version comparison */
+export interface VersionComparison {
+    baseVersion: string;
+    compareVersion: string;
+    changes: VersionChange[];
+    statistics: VersionStatistics;
+}
+/** Version change */
+export interface VersionChange {
+    type: 'added' | 'modified' | 'deleted';
+    path: string;
+    sampleId?: string;
+    details?: ChangeDetails;
+}
+/** Change details */
+export interface ChangeDetails {
+    before?: unknown;
+    after?: unknown;
+    diff?: unknown;
+}
+/** Version statistics */
+export interface VersionStatistics {
+    added: number;
+    modified: number;
+    deleted: number;
+    sizeDelta?: number;
+}
+/** Data quality report */
+export interface DataQualityReport {
+    datasetId: string;
+    timestamp: Date;
+    overall: QualityScore;
+    dimensions: QualityDimension[];
+    issues: QualityIssue[];
+    recommendations?: QualityRecommendation[];
+}
+/** Quality score */
+export interface QualityScore {
+    score: number;
+    grade: 'A' | 'B' | 'C' | 'D' | 'F';
+}
+/** Quality dimension */
+export interface QualityDimension {
+    name: string;
+    score: number;
+    weight?: number;
+    details?: string;
+}
+/** Quality issue */
+export interface QualityIssue {
+    id: string;
+    type: QualityIssueType;
+    severity: 'low' | 'medium' | 'high' | 'critical';
+    description: string;
+    affectedSamples?: number;
+    column?: string;
+    examples?: unknown[];
+}
+/** Quality issue type */
+export type QualityIssueType = 'missing_values' | 'duplicates' | 'outliers' | 'inconsistent_format' | 'invalid_values' | 'class_imbalance' | 'label_noise' | 'data_drift' | 'schema_mismatch' | 'custom';
+/** Quality recommendation */
+export interface QualityRecommendation {
+    issue: string;
+    action: string;
+    impact: 'low' | 'medium' | 'high';
+    automated?: boolean;
+}
+/** Export task */
+export interface ExportTask {
+    id: string;
+    datasetId: string;
+    format: ExportFormat;
+    options: ExportOptions;
+    status: ExportStatus;
+    result?: ExportResult;
+    error?: string;
+    startedAt?: Date;
+    completedAt?: Date;
+}
+/** Export format */
+export type ExportFormat = 'csv' | 'json' | 'jsonl' | 'parquet' | 'coco' | 'yolo' | 'voc' | 'labelme' | 'tfrecord' | 'huggingface' | 'custom';
+/** Export options */
+export interface ExportOptions {
+    split?: DataSplit[];
+    columns?: string[];
+    includeLabels?: boolean;
+    labelFormat?: string;
+    includeMedia?: boolean;
+    compression?: 'none' | 'gzip' | 'zip';
+    partitionBy?: string;
+    maxFileSize?: number;
+}
+/** Export status */
+export type ExportStatus = 'pending' | 'processing' | 'completed' | 'failed' | 'cancelled';
+/** Export result */
+export interface ExportResult {
+    files: ExportFile[];
+    totalSize: number;
+    rowCount: number;
+    downloadUrl?: string;
+    expiresAt?: Date;
+}
+/** Export file */
+export interface ExportFile {
+    name: string;
+    path: string;
+    size: number;
+    format: string;
+    checksum?: string;
+}
+//# sourceMappingURL=dataset-management.d.ts.map