Skip to content

众包集成

与 Prolific、MTurk 和其他众包平台进行集成。

众包集成

Potato 与 Prolific 和 Amazon Mechanical Turk 等众包平台无缝集成，用于大规模标注任务。

Prolific 集成

基本设置

yaml
crowdsourcing:
  platform: prolific
  enabled: true
  completion_code: "POTATO2024"  # Code shown on completion

URL 参数

Prolific 通过 URL 参数传递参与者信息：

yaml
crowdsourcing:
  platform: prolific
  url_params:
    - PROLIFIC_PID    # Participant ID
    - STUDY_ID        # Study ID
    - SESSION_ID      # Session ID

工作者通过以下方式访问：

text
https://your-server.com/?PROLIFIC_PID=xxx&STUDY_ID=xxx&SESSION_ID=xxx

Prolific 配置

在你的 Prolific 研究设置中：

  1. 将研究 URL设置为你的 Potato 服务器
  2. 添加 URL 参数：?PROLIFIC_PID={{%PROLIFIC_PID%}}&STUDY_ID={{%STUDY_ID%}}&SESSION_ID={{%SESSION_ID%}}
  3. 将完成代码设置为与你的配置匹配

验证

验证 Prolific 参与者：

yaml
crowdsourcing:
  platform: prolific
  validate_participant: true
  completion_code: "POTATO2024"

Amazon MTurk 集成

基本设置

yaml
crowdsourcing:
  platform: mturk
  enabled: true

HIT 配置

创建指向你服务器的外部问题 HIT：

xml
<?xml version="1.0" encoding="UTF-8"?>
<ExternalQuestion xmlns="http://mechanicalturk.amazonaws.com/AWSMechanicalTurkDataSchemas/2006-07-14/ExternalQuestion.xsd">
  <ExternalURL>https://your-server.com/?workerId=${workerId}&amp;assignmentId=${assignmentId}&amp;hitId=${hitId}</ExternalURL>
  <FrameHeight>800</FrameHeight>
</ExternalQuestion>

URL 参数

yaml
crowdsourcing:
  platform: mturk
  url_params:
    - workerId
    - assignmentId
    - hitId

沙盒测试

先用 MTurk 沙盒进行测试：

yaml
crowdsourcing:
  platform: mturk
  sandbox: true  # Use sandbox environment

工作者管理

追踪工作者

yaml
crowdsourcing:
  track_workers: true
  worker_id_field: worker_id

限制每个工作者的实例数

yaml
instances_per_annotator: 50

阻止重复工作者

防止工作者重复参与任务：

yaml
crowdsourcing:
  prevent_retakes: true

质量控制

注意力检查

插入测试问题：

yaml
attention_checks:
  enabled: true
  frequency: 10  # Every 10 instances
  fail_threshold: 2
  action: warn  # or 'block'

金标准问题

json
{
  "id": "gold_1",
  "text": "The sky is typically blue during a clear day.",
  "gold_label": "True",
  "is_gold": true
}
yaml
quality_control:
  gold_questions: true
  gold_percentage: 10  # 10% of instances
  min_gold_accuracy: 70

时间限制

yaml
crowdsourcing:
  min_time_per_instance: 5  # seconds
  max_time_total: 3600  # 1 hour

拒绝低质量工作

yaml
quality_control:
  auto_reject:
    enabled: true
    conditions:
      - gold_accuracy_below: 50
      - completion_time_under: 300  # seconds

完成处理

显示完成代码

yaml
completion:
  show_code: true
  code: "POTATO2024"
  message: "Thank you! Your completion code is: {code}"

完成后重定向

yaml
completion:
  redirect: true
  redirect_url: "https://prolific.co/submissions/complete?cc={code}"

自定义完成页面

yaml
completion:
  custom_template: templates/completion.html

支付等级

基于质量

yaml
payment:
  tiers:
    - name: bonus
      condition:
        gold_accuracy_above: 90
      amount: 0.50
    - name: standard
      condition:
        gold_accuracy_above: 70
      amount: 0.00
    - name: reject
      condition:
        gold_accuracy_below: 50

完整示例：Prolific 研究

yaml
task_name: "Sentiment Analysis Study"
 
# Crowdsourcing settings
crowdsourcing:
  platform: prolific
  enabled: true
  completion_code: "SENT2024"
  url_params:
    - PROLIFIC_PID
    - STUDY_ID
    - SESSION_ID
  prevent_retakes: true
 
# Open access for crowdworkers
allow_all_users: true
 
# Task assignment
instances_per_annotator: 50
annotation_per_instance: 3
 
# Quality control
attention_checks:
  enabled: true
  frequency: 10
  fail_threshold: 2
 
quality_control:
  gold_questions: true
  gold_percentage: 5
  min_gold_accuracy: 70
 
# Data
data_files:
  - path: data/main.json
    text_field: text
 
# Annotation scheme
annotation_schemes:
  - annotation_type: radio
    name: sentiment
    description: "What is the sentiment?"
    labels:
      - Positive
      - Negative
      - Neutral
    keyboard_shortcuts:
      Positive: "1"
      Negative: "2"
      Neutral: "3"
 
# Completion
completion:
  show_code: true
  code: "SENT2024"
  message: |
    ## Thank you for participating!
 
    Your completion code is: **{code}**
 
    Please return to Prolific and enter this code to receive payment.

完整示例：MTurk HIT

yaml
task_name: "Image Classification HIT"
 
crowdsourcing:
  platform: mturk
  enabled: true
  url_params:
    - workerId
    - assignmentId
    - hitId
 
allow_all_users: true
instances_per_annotator: 20
 
# Time constraints
crowdsourcing:
  min_time_per_instance: 3
  max_time_total: 1800
 
# MTurk form submission
completion:
  mturk_submit: true
  submit_url: "https://www.mturk.com/mturk/externalSubmit"
 
annotation_schemes:
  - annotation_type: radio
    name: category
    description: "What is shown in this image?"
    labels:
      - Cat
      - Dog
      - Bird
      - Other

监控工作者

管理员仪表板

yaml
admin_users:
  - researcher@university.edu
 
admin_dashboard:
  enabled: true
  show_worker_stats: true

访问 /admin 可查看：

  • 工作者完成率
  • 每个实例的平均时间
  • 金标准准确率
  • 注意力检查结果

导出工作者数据

bash
potato export-workers config.yaml --output workers.csv

最佳实践

  1. 充分测试 - 先用小组进行试点
  2. 设定公平薪酬 - 计算预估时间并公平支付
  3. 清晰的说明 - 包含示例和边界情况
  4. 使用注意力检查 - 捕获随机点击行为
  5. 包含金标准问题 - 验证理解程度
  6. 实时监控 - 尽早发现问题
  7. 计划拒绝策略 - 预先设定明确的质量标准
  8. 沟通问题 - 就问题与工作者联系
  9. 根据反馈迭代 - 根据工作者意见进行改进
  10. 定期导出数据 - 不要等到最后

延伸阅读

有关实现细节，请参阅源代码文档。