一、岗位与角色
– 人工智能训练师｜AI Trainer
– 数据标注员｜Data Annotator
– 数据标注工程师｜Data Annotation Engineer
– 算法工程师｜Algorithm Engineer
– 数据科学家｜Data Scientist
– 项目经理｜Project Manager
– 质检专员｜Quality Control Specialist
– 标注组长｜Annotation Team Leader
– 模型运营｜Model Operation
– prompt 工程师｜Prompt Engineer
二、数据与标注基础
– 数据集｜Dataset
– 训练数据｜Training Data
– 测试数据｜Test Data
– 验证数据｜Validation Data
– 语料｜Corpus
– 样本｜Sample
– 标注｜Annotation
– 标注规范｜Annotation Guideline
– 标注工具｜Annotation Tool
– 数据清洗｜Data Cleaning
– 数据预处理｜Data Preprocessing
– 数据脱敏｜Data Desensitization
– 数据去重｜Data Deduplication
– 数据增强｜Data Augmentation
– 结构化数据｜Structured Data
– 非结构化数据｜Unstructured Data
– 半结构化数据｜Semi-structured Data
– 多轮对话｜Multi-turn Dialogue
– 单轮对话｜Single-turn Dialogue
三、标注任务类型
– 文本分类｜Text Classification
– 实体标注｜Entity Annotation
– 命名实体识别｜Named Entity Recognition (NER)
– 关系抽取｜Relation Extraction
– 意图识别｜Intent Recognition
– 情感分析｜Sentiment Analysis
– 观点抽取｜Opinion Mining
– 文本纠错｜Text Correction
– 文本摘要｜Text Summarization
– 机器翻译校对｜MT Post-editing
– 问答对构建｜QA Pair Construction
– 多轮续写｜Dialogue Continuation
– 安全审核｜Safety Review
– 有害内容标注｜Harmful Content Annotation
– 偏好标注｜Preference Annotation
– 排序标注｜Ranking Annotation
– 对比标注｜Comparison Annotation
四、模型训练与优化
– 模型训练｜Model Training
– 模型微调｜Fine-tuning
– 监督学习｜Supervised Learning
– 无监督学习｜Unsupervised Learning
– 强化学习｜Reinforcement Learning
– 人类反馈强化学习｜RLHF
– 模型迭代｜Model Iteration
– 模型效果｜Model Performance
– 泛化能力｜Generalization
– 过拟合｜Overfitting
– 欠拟合｜Underfitting
– 收敛｜Convergence
– 推理｜Inference
– 部署｜Deployment
五、质量与评估
– 准确率｜Accuracy
– 准确率｜Precision
– 召回率｜Recall
– F1值｜F1 Score
– 一致性｜Consistency
– 标注准确率｜Annotation Accuracy
– 返工｜Rework
– 抽检｜Spot Check
– 全检｜Full Inspection
– 误标｜Misannotation
– 漏标｜Missed Annotation
– 歧义｜Ambiguity
– 边界案例｜Edge Case
– bad case｜Bad Case
– 效果评估｜Effect Evaluation
– 满意度｜Satisfaction
六、提示词与交互
– 提示词｜Prompt
– 指令｜Instruction
– 系统提示｜System Prompt
– 角色设定｜Role Setting
– 约束条件｜Constraint
– 输出格式｜Output Format
– 上下文｜Context
– 对话历史｜Conversation History
– 生成内容｜Generated Content
– 回复优化｜Response Optimization
– 对齐｜Alignment
– 人类偏好｜Human Preference
七、安全与伦理
– 内容安全｜Content Safety
– 隐私保护｜Privacy Protection
– 合规性｜Compliance
– 偏见｜Bias
– 歧视｜Discrimination
– 有害信息｜Harmful Information
– 暴力内容｜Violent Content
– 色情内容｜Pornographic Content
– 虚假信息｜Misinformation
– 伦理规范｜Ethical Guidelines
八、项目与流程
– 需求分析｜Requirement Analysis
– 任务拆解｜Task Breakdown
– 标注流程｜Annotation Process
– 交付标准｜Delivery Standard
– 工期｜Schedule
– 效率｜Efficiency
– 产能｜Capacity
– 复盘｜Review
– 优化提议｜Optimization Suggestions
© 版权声明
文章版权归作者所有，未经允许请勿转载。
相关文章
暂无评论...
