SBIR Phase I: Adaptive Dialog Systems As Second Language Learning Partners

这个小企业创新研究（SBIR）第一阶段项目的广泛影响集中在提供一种成本更低、更容易获得的替代方案，帮助世界各地的人们学习说一门新语言。通过使用最先进的对话式人工智能（AI），该项目为学生提供个性化的、基于课程的学习，这种学习每天24小时都可以负担得起，而且没有时间限制。此外，人工智能系统将能够分析学习者的语音，并就如何提高语法和会话流畅性提供建议，就像训练有素的导师帮助学生一样。该项目将会话式人工智能引入语言学习市场，使其成为一种更有效、更有趣、更低成本的产品，从而改变当前的语言学习模式。这项小企业创新研究（SBIR）第一阶段项目是一项创新整合工作，旨在将最先进的人工智能技术结合到教育应用中。该项目将为语言学习任务的对话策略培训提供新的用户自适应算法和数据集。该项目将为公司的开源、多模式对话系统框架做出贡献，使对话社区能够减少多模式对话系统研究的进入壁垒。该项目还将为社区提供一个带注释的会话数据集，用于语法错误纠正，以及训练有素的语法纠正模型和语法错误类型检测器。发展起来的第二语言交际学习的教育对话系统也可以作为教育专家检验不同学习理论的一般框架。本项目还将探讨适应性会话经验和不同反馈结构是否会影响学习者信心和学习收益的假设。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact of this Small Business Innovation Research (SBIR) Phase I project centers on offering a lower-cost, more accessible alternative to help people everywhere learn to speak a new language. By using state-of-the-art conversational artificial intelligence (AI), the project offers students individualized, curriculum-based learning that is affordable and accessible 24 hours a day, with no limits on the amount of time they can practice. Additionally, the AI system will be able to analyze learners’ speech to offer suggestions on ways to improve their grammar and conversational fluency, in much the same way a trained tutor would assist a student. The project brings conversational AI to the language learning market, enabling a more effective, entertaining, and low-cost product to transform the current language learning paradigm.This Small Business Innovation Research (SBIR) Phase I project is an innovative integration effort to combine state-of-the-art AI technologies for education applications. The project will contribute new user-adaptive algorithms and datasets for dialog policy training on language learning tasks. The project will contribute to the company’s open-source, multimodal, dialog system framework to enable the dialog community to reduce the entry barrier for multimodal dialog system research. The project will also provide the community with an annotated conversational dataset for grammatical error correction along with a trained grammar correction model and grammar error type detector. The developed educational dialog systems for second language communication learning can also serve as a general framework for education experts to test different learning theories. This project will also explore hypotheses of whether adaptive conversation experience and different feedback structures will impact learner confidence and learning gain.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.