Using Intelligent Tutors to Teach Students How APIs Are Used for Software Engineering in Practice

Abstract

Abstract-Computer science courses typically incorporate integrated training in software engineering, which includes learning how to reuse existing code libraries in new programs. This commonly presents a need to use the libraries' application programming interfaces (APIs) during project-based learning activities. The students learn these APIs with little support from the instructor. Instead, they primarily learn from whatever code examples they find on the web-a scattershot approach that is slow, frustrating and sometimes fruitless. Preliminary work investigated an improved approach that provides students with intelligent API tutors. This interactive educational content walks students through code examples selectively harvested from the web. This approach increased students' knowledge of API method names, but a key pedagogical challenge remained: students still lacked an understanding of common problems and surprises that they might encounter when using APIs in practice. Perhaps even more crucially, the prior statistical study provided no information about how students would perceive and accept the intelligent tutor approach, when given a choice. Therefore, the current paper presents an enhanced approach that augments intelligent tutors with selected information from online FAQs and online open source code, thereby providing more explanation and context about how to use APIs in practice. A qualitative study assessed student perceptions of the system's usability as well as suggestions for improvement, revealing that 100% of students considered the approach to be useful. A key implication is that providing information from online FAQs, related articles, and open source code could facilitate computer science students' learning how to use APIs.

A Cognitive Assistant that Uses Small Talk in Tutoring Conversation

“…In this work, we preferred the term "cognitive assistant", because the system not only returns answers for a given question, but also learns from its interactions with the students. Many research publications in this field provide a set of denominations other than "cognitive assistant" [24], [25] or "cognitive tutor" [26], but we can also find terms such as "chatbot assistant" [27], "personal assistant" [5], [28], [29], [30], [31], [32], "intelligent tutors" [33], [34], [35], [36], "knowledge exploration assistant" [37], "ontological learning assistant" [38] and "conversational agent" [39][40] [41]. With respect to small talk, one relevant work is by Bickmore and Cassel [6], who described the development of an embodied conversational interface agent capable of understanding and generating multimodal inputs and outputs, operating on a limited application domain in which both social and task-oriented dialogue are important.…”

“…Krishnamoorthy et al [33] presented an enhanced approach that augments intelligent tutors with selected information from online FAQs (Frequently Asked Questions) and online open-source code, thereby providing more explanation and context about how to use APIs (Application Programming Interfaces) in practice.…”

Intelligent tutoring Module for a 3Dgame-based science e-learning platform

“…With decades of research and development, ITS now have been built for various scientific domains such as physics [36], [22], [19], mathematics [9], [1], [41], chemistry [3], computer programming [16], [42], engineering [7], [26], medicine [13], [40], and so on. These studies showed that virtual manipulative through the ITS can significantly improve the learning.…”