Affiliation:
1. University of Southern California, Admiralty Way, Marina del Rey CA, USA
Abstract
Humans learn procedures from one another through a variety of methods, such as observing someone do the task, practicing by themselves, reading manuals or textbooks, or getting instruction from a teacher. Some of these methods generate examples that require the learner to generalize appropriately. When procedures are complex, however, it becomes unmanageable to induce the procedures from examples alone. An alternative and very common method for teaching procedures is tutorial instruction, where a teacher describes in general terms what actions to perform and possibly includes explanations of the rationale for the actions. This article provides an overview of the challenges in using human tutorial instruction for teaching procedures to computers. First, procedures can be very complex and can involve many different types of interrelated information, including (1) situating the instruction in the context of relevant objects and their properties, (2) describing the steps involved, (3) specifying the organization of the procedure in terms of relationships among steps and substeps, and (4) conveying control structures. Second, human tutorial instruction is naturally plagued with omissions, oversights, unintentional inconsistencies, errors, and simply poor design. The article presents a survey of work from the literature that highlights the nature of these challenges and illustrates them with numerous examples of instruction in many domains. Major research challenges in this area are highlighted, including the difficulty of the learning task when procedures are complex, the need to overcome omissions and errors in the instruction, the design of a natural user interface to specify procedures, the management of the interaction of a human with a learning system, and the combination of tutorial instruction with other teaching modalities.
Funder
Defense Advanced Research Projects Agency
National Science Foundation
Air Force Office of Scientific Research
Publisher
Association for Computing Machinery (ACM)
Subject
Artificial Intelligence,Human-Computer Interaction
Reference104 articles.
1. The future of end user programming?
2. Interaction, Comprehension, and Instruction Usage
3. C. Anderson. 2008. The Long Tail: Why the Future of Business Is Selling Less of More. Hyperion. C. Anderson. 2008. The Long Tail: Why the Future of Business Is Selling Less of More. Hyperion.
4. Natural language interfaces to databases – an introduction
Cited by
4 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献
1. Unpacking Human Teachers’ Intentions for Natural Interactive Task Learning;2021 30th IEEE International Conference on Robot & Human Interactive Communication (RO-MAN);2021-08-08
2. Enriching how-to guides with actionable phrases and linked data;Web Intelligence;2017-08-11
3. Towards Automating Data Narratives;Proceedings of the 22nd International Conference on Intelligent User Interfaces;2017-03-07
4. Enriching How-to Guides by Linking Actionable Phrases;Proceedings of the 25th International Conference Companion on World Wide Web - WWW '16 Companion;2016