Automatic Generation of Procedural Knowledge Using Program Synthesis
user-5ebe28444c775eda72abcdcf(2019)
摘要
One of the driving applications of technology has been augmenting what humans can do and how they approach solving problems. Specifically within design fields, computer-powered tools (eg, Computer-Aided Design for architecture/industrial design, Photoshop/Illustrator for visual art) have revolutionized the way people approach design problems, changed what is possible, and expanded how many people can fruitfully work on such problems. Computers can help designers by taking care of rote tasks and computation, freeing designers’ time and brainpower to focus on the challenging aspects of problems that cannot be automated. Beyond serving simply as a tool, computers can actually contribute the design, working with the designer to solve problems. This thesis focuses on using computers to aid in the particular class of design tasks involving models of human expertise and learning in problem domains such as high school mathematics or puzzles. A motivating example of this class of tasks is that of building educational software for a domain such as solving linear algebraic equations. A key challenge in the design of educational applications is modeling the operational knowledge that represents the expertise and skill of human problem solving. Designing and building this model—called the domain model—is necessary and challenging for any educational application. For example, one needs to decide what should be taught in a class curricula. However, a computer-based tutor needs a level of formality and detail in the domain model that makes this design task even more difficult. Solving the domain-model design task lies at the heart of …
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络