Recently, AIED 2021 (22nd International Conference on Artificial Intelligence in Education), the top international conference on artificial intelligence education, opened grandly. With the outstanding achievements of publishing 5 academic papers, the future has demonstrated the development potential of China’s AI+ education to the world. At the meeting, the author of the thesis focused on the cutting-edge research results of the Future AI Research Institute in the fields of NLP (Natural Language Processing) and multimodal learning, and conducted in-depth academic discussions and technical exchanges with experts and scholars from all over the world.
AIED is the top international conference in the field of educational applications. It is organized by the International Association for Artificial Intelligence Education (IAIED). It is famous for “providing high-quality research intelligent systems and cognitive scientific methods for the field of educational computing applications”. The included papers are representative The latest development direction and level of artificial intelligence in the field of education applications. The five papers included this time are based on the results of NLP and multi-modal learning research based on real application scenarios in the future, and are implemented in specific businesses, covering composition correction, automatic problem solving, teaching behavior analysis, teaching quality evaluation, and teacher Personalized matching and other topics.
Breakthroughs in the field of NLP and multimodal learning, Implement multiple teaching scenarios
There is a huge difference between the teaching scene of online education and traditional offline education. Many teachers are accustomed to face-to-face teaching to students in offline classrooms. When facing the cameras of computers and other equipment, they are often inexperienced and it is difficult to fully control the classroom. They often miss or miss the best time to issue teaching instructions, such as encouraging students, Remind students to take notes, etc. Good Future AI Research Institute hopes to use an automated system to capture important teaching instructions and provide feedback to help improve teachers’ online teaching level. This idea is implemented in the “Multi-Task Learning based Online Dialogic Instruction Detection with Pre-trained Language Models” included in AIED2021. The teaching proposed in the article The instruction detection method has been launched and applied to the GodEye classroom quality detection system developed by Haowei to provide detection services for online teaching behaviors.
In the GodEye system, the performance of teachers and students in the classroom can be presented digitally, the overall course quality, the teaching situation of teachers and the learning state of students can be quantified, the classroom can be disassembled with data indicators, and the full-link management of the online classroom can be realized.
GodEye classroom quality inspection solution system structure GodEye has been widely used in Xueersi one-to-one business. During the 2020 epidemic, it will cumulatively analyze more than 4.6 million hours of online courses to provide underlying data support for improving the quality of teaching. After connecting to the system, the situation of teachers making timely teaching instructions in online courses has been improved. Taking the important teaching instruction “remind students to take notes” as an example, sampling analysis shows that as the number of days of use increases, teachers remind students to take notes. The number of times has gradually increased, effectively ensuring the quality of online teaching. The research results have made many firsts, accurately matching the needs of teachers and students It is reported that Good Future’s papers selected for this conference have created multiple industry firsts: for the first time, the teacher-student matching model under the online one-to-one teaching mode is modeled, and the history of class information and class behavior of analysts is used for each student. Match the right teacher accurately. With the blessing of AI technology in the future, it will be possible to achieve precise matching of teachers and students in the future, and customize personalized courses with “thousands of people and thousands of faces”, accelerating the arrival of the era of teaching students in accordance with their aptitude. For the first time, Good Future also proposed an automatic evaluation method for key writing based on machine reading comprehension. Through machine reading, it can comprehend the title and composition, output the writing score, and can also display the key points in the student’s composition. In this way, students can understand which way of writing is correct in order to maintain good writing habits. At the same time, this method can also tell students which points are not responded to in the composition, help students avoid problems and improve methods, and improve writing skills and abilities. The traditional automatic scoring method can only provide a total score at the content level. Detailed special feedback requires the introduction of manual labor. In comparison, the new technology of the future can provide users with more feedback information and accelerate the feedback process. The research results have been used in the automatic composition scoring of English composition correction projects. Five years of sharpening a sword, good future AI achievements are outstanding Good future’s research results in the field of AI research stem from its understanding and attention to AI+ education. In recent years, Good Future has continuously increased its investment in AI research and development. At present, it has focused on the needs of education scenarios. It has accumulated research and development including 8 types of images, voice, data mining, natural language processing, etc., and more than 100 AI capabilities, creating more than 10 items. AI solutions are used in education scenarios, covering all teaching links of “preparation, teaching, practice, examination, evaluation, and management”. At present, Good Future has also commercialized a number of AI capabilities and widely used them in a number of internal businesses. For example, Good Future’s Chinese and English composition correction program can not only assist in checking and correcting grammatical errors, and help teachers improve the efficiency of correction, but also help students check the composition and quickly give the result of composition correction, which has become an industry benchmark. Introduction to Chinese and English composition correction function Good Future attaches great importance to technology research and development and the construction of the underlying scientific research capabilities, while not forgetting the original intention of “making education better”. It has established close industry-university-research cooperation with Tsinghua University, the Institute of Computing Technology of the Chinese Academy of Sciences and other famous universities and Dozens of academic achievements of the students were selected into top international academic conferences such as ICASSP, NeurIPS, AAAI, WWW, EMNLP, AIED, NCME, etc. In the launch event of the Beijing National Artificial Intelligence Innovation Application Pilot Zone that just ended, Good Future served as the vice chairman unit of the Beijing Artificial Intelligence Industry Alliance, and as the only vice chairman unit from the education technology industry was invited to attend, shouldering the responsibility of advancing artificial intelligence. Industrial coordination and ecological construction will fully support the sustainable and healthy development of Beijing’s artificial intelligence industry. As the contractor of a new generation of artificial intelligence open innovation platform in the smart education country, Good Future will continue to adhere to the principles of source innovation, application-driven, and open sharing, assume the responsibility of promoting the integrated development of AI + education, and work with industry partners to help the education industry to become intelligent upgrade