Design And Implementation Of Assistant Tool For Teaching Speech Act Analysis In Teaching Video

In recent years,with the rapid development of artificial intelligence technology and its deep integration with education,it has greatly promoted the development of education and teaching in our country.The use of intelligent means to analyze and research teaching has become a research hotspot.Among them,the actual classroom is the main front of the current education and teaching research.Research on classroom teaching behavior is of great significance for discovering the law of classroom teaching and promoting the professional development of teachers.Teaching behavior can be divided into speech act and nonverbal act,among which speech act is the main way of teacher-student interaction in the classroom,accounting for about 80%of all teaching behaviors[1].However,there is very little research on intelligent analysis of teaching speech act in the classroom.Some scholars have used text classification algorithms to conduct research on teaching speech act,but they involve few categories and ignore the contextual relevance of the teaching speech acts in the time sequence.In order to help teaching researchers to analyze the teaching speech act in the classroom more conveniently,provide an auxiliary tool for teaching researchers,reduce the huge workload caused by manual sorting and coding,and solve the problem of inability to understand the meaning of teaching at a deeper level caused by the lack of relevant contextual information in the existing intelligent analysis methods of teaching speech act.This paper takes the teaching speech act text from actual recording teaching video in the classroom as the research object,uses the text analysis method that integrates context information to study the automatic recognition of the teaching speech act in the teaching video,and develops the corresponding analysis assistant tools.The specific research contents are as follows:(1)Compilation of coding table of teaching speech act analysis in teaching video.According to the classification of teaching speech act in Flanders Interactive Analysis System,combined with the characteristics of teaching speech act in teaching video.The coding table of teaching speech act analysis in teaching video is proposed,which provides a theoretical basis for follow-up research.(2)Research on teaching speech act recognition algorithm in teaching videos.This paper regards the problem of teaching speech act recognition as a sequence labeling problem,and proposes a teaching speech act recognition algorithm that integrates contextual information.In this algorithm,the Bi-directional LSTM network based on attention mechanism is used to encode sentences.Then,the Bi-directional LSTM is used to fuse the contextual information of teaching speech act at the teaching speech sequence level,and the CRF layer is added to mine the implicit contextual relationship in the label layer.Through the use of the teaching speech act corpus in the teaching video constructed in this paper,the teaching speech act recognition model fused with contextual information is trained and tested.The effectiveness of the proposed algorithm is verified.(3)Based on the research on the teaching speech act recognition algorithm in the teaching video,the auxiliary tool for teaching speech act analysis in the teaching video is designed and implemented.It includes teaching video speech transcription module,teaching speech act recognition module,teaching speech act analysis module,teaching speech content analysis module,etc.The aim is to use intelligent technology to help teaching researchers to analyze the teaching speech act in teaching video.