Inducing programs in a direct-manipulation environment
CHI '89 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
EAGER: programming repetitive tasks by example
CHI '91 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
The computer user as toolsmith: the use, reuse, and organization of computer-based tools
The computer user as toolsmith: the use, reuse, and organization of computer-based tools
C4.5: programs for machine learning
C4.5: programs for machine learning
Watch what I do: programming by demonstration
Watch what I do: programming by demonstration
Agents that reduce work and information overload
Communications of the ACM
APE: learning user's habits to automate repetitive tasks
Proceedings of the 5th international conference on Intelligent user interfaces
Programming by example: novice programming comes of age
Communications of the ACM
Machine Learning
A Comparison of Decision Tree Ensemble Creation Techniques
IEEE Transactions on Pattern Analysis and Machine Intelligence
JavaScript: The Definitive Guide
JavaScript: The Definitive Guide
CoScripter: automating & sharing how-to knowledge in the enterprise
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
Toward a programming laboratory
IJCAI'69 Proceedings of the 1st international joint conference on Artificial intelligence
Lowering the barriers to website testing with CoTester
Proceedings of the 15th international conference on Intelligent user interfaces
| Hi-index | 0.00 |
Web applications such as web-based email, spreadsheets and form filling applications have become ubiquitous. However, many of the tasks that users try to accomplish with such web applications are highly repetitive. In this paper we present the design of a system we have developed that learns and thereafter automates users' repetitive tasks in web applications. Our system infers users' intentions using an ensemble of decision trees. This enables it to handle branching, generalization and recurrent changes of relative and absolute positions. Our evaluation shows that our system converges to the correct solution after 3--8 iterations when the pattern is noise-free, and after 3--14 iterations for a noise level between 5--35%.