Crowd-Powered Data Mining
Information.
KDD 2018 Website
Time: 1:00 PM - 5:00 PM
Location: ICC Capital Suite Room 17 (Level 3)
Slides
Abstract.
Many data mining tasks cannot be completely addressed by automated processes, such as sentiment analysis and image classification. Crowdsourcing is an effective way to harness the human cognitive ability to process these machine-hard tasks. Thanks to public crowdsourcing platforms, e.g., Amazon Mechanical Turk and CrowdFlower, we can easily involve hundreds of thousands of ordinnary workers (i.e., the crowd) to address these machine-hard tasks. In this tutorial, we will survey and synthesize a wide spectrum of existing studies on crowd-powered data mining. We first give an overview of crowdsourcing, and then summarize the fundamental techniques, including quality control, cost control, and latency control, which must be considered in crowdsourced data mining. Next we review crowd-powered data mining operations, including classification, clustering, pattern mining, machine learning using the crowd (including deep learning, tansfer learning and semi-supervised learning) and knowledge discovery. Finally, we provide the emerging challenges in crowdsourced data mining.
Outline.
1. Crowdsourcing Overview (20 minutes)
- Crowdsourcing motivation
- Crowdsourcing workflow
- Crowdsourcing applications
- Crowdsourcing platforms
2. Quality control (40 minutes)
- Characterize a worker's quality
- Eliminate the low-quality workers
- Truth inference
- Task Assignment
3. Cost control (40 minutes)
- Machine-based pruning
- Task Selection
- Answer deducation
- Sampling
- Task design
Break (30 minutes)
4. Latency control (20 minutes)
- Single-task latency control
- Single-batch latency control
- Multi-batch latency control
5. Crowd Mining (60 minutes)
- Crowd-powered Pattern mining (10 minutes)
- Crowd-powered Classification (10 minutes)
- Crowd-powered Clustering (10 minutes)
- Crowd-powered Machine Learning (10 minutes)
- Deep Learning
- Transfer Learning
- Semi-supervised Learning
- Crowd-powered Knowledge Discovery (20 minutes)
6. Challenges (10 minutes)
