Harvesting the great power of the human mind in the most efficient and reliable manner is a topic that has motivated scientists for many decades now, even before the existence of computer systems. This module will provide an overview of the concepts of Human Computation and Crowdsourcing. The main focus of the module will be to provide participants with the tools and methodologies to tap into the power of the crowd, especially when this crowd can be reached over the Internet and with the help of computers and smart devices. 

HCI612 aspires to be a valuable module to HCI specialists who work with Machine Learning and AI practitioners, to a more general audience that interacts on an every-day basis with AI systems and even for social psychology experts interested in the behavior of humans participating in crowdsourcing processes. Apart from a brief overview of the main concepts and a short historic overview of the terms, the course aspires to exhibit the vast influence of human computation and crowdsourcing into society and especially in the creation of machine learning datasets. Additionally, the module aspires to educate the participants on the methodological pitfalls and parameters, responsible for low quality results in crowdsourcing and biases in the created data.  With the successful compilation of the module participants will gain awareness on this ever-growing field but most importantly will be capable of generating their own high-quality datasets through crowdsourcing in an efficient, reliable and bias-aware manner. 

In particular, throughout the duration of the course, students of HCI612 will have the opportunity to:

1. Understand the role of human computation and crowdsourcing in the development of AI systems and machine learning datasets. Special emphasis will be given in data labeling generation for computer vision applications. 
2. Observe the benefits and challenges of human computation and crowdsourcing. 
3. View the main actors in a crowdsourcing process and analyze their motives.   
4. Learn best practices for designing a crowdsourcing task to ensure data quality and validity.
5. Understand the social bias inherent in the produced data and ways to overcome this obstacle.
6. Understand the ethical considerations of the approach and possible shortcomings or subjects difficult to address. 

Apart from a deep understanding of the theory on human computation and crowdsourcing topics, participants are expected to also gain some practical experience with the concept of crowdsourcing. It is envisioned that participants will be able to operate a micro-task crowdsourcing platform for a given project’s needs, as well as to collect and analyze the created dataset.

Module content

1. Introduction to the course
2. Introduction to Human Computation 
3. Crowdsourcing in everyday life 
4. Main actors in Crowdsourcing and Human Computation 
5. Designing a crowdsourcing task 
6. Requesters: How to achieve high quality task results
7. Data generation for Machine Learning I
8. Data generation for Machine Learning II: Data cleaning and Analysis 
9. Social Bias in Crowdsourced Datasets I
10. Social Bias in Crowdsourced Datasets II
11. Challenges in Crowdsourcing: Achieving Diversity; Reproducible and Dynamic Datasets
12. Advanced topics