Providing Tutoring Service through Accumulating Interaction Data ChiJen

Providing Tutoring Service through Accumulating Interaction Data Chi-Jen LIN Fo Guang University, Taiwan

The Needs of Tutoring n Elementary schools, high schools, and universities n especially for low learning achievement students Homework n Classroom teaching n

Providing Tutoring Service = AI ? n Knowledge models in an typical ITS Student model n Domain model n Pedagogical model n AI / knowledge engineering background required: unsuitable for ordinary teaching n

Human Tutoring n Most domain experts are effective tutors (Chi, et al. , 2001) Domain knowledge required n Tutoring skills unimportant n Accumulating tutoring knowledge through tutoring n

Tutoring Service in Focus Reducing human tutoring load by removing repeated parts n Tutoring service without knowledge model development n Accumulating human tutoring interaction data n

Tutoring Interaction Data Model n Interaction session n n Episode n n Tutor-student interaction on a specific topic or task A student action (situation) + a tutor action (response) Situation: n A student action to be responded by the tutor n n Ex: an input of a student step in problem solving, a student question Response: n The response given by the tutor n could be nothing

Database of Tutoring Interaction

Accumulating Tutoring Interaction Data Student interface Student Teacher interface Teacher action Tutoring support Student action Backend server Help request Teacher

Interaction Data Reuse: Matching n Perfect matching n n Problem-solving path matching n n All situation and response in the session matched All student steps in the session matched Current situation matching n Current student action matched

Application 1: Linear Equation Exercises To familiarize students with techniques in solving linear equations n To correct common student mistakes by providing feedbacks to student errors n

Data Repetition Problem: 3 x – 4 = 2 x -1 2[3(2 x-5)-1] = 7 x+3 (2 x+5)/4 -3(x 1/2)=1 Student step repetition 93% 87% 66%

Application 2: Learning of Syntax and Semantic of Computer Languages n Students tend to remember the surface meaning of program codes without actually understand their semantic n n eg: boy_mc. x = boy_mc. x + 5; Provide students with code interpretation exercises n computer feedbacks to student answers

Conclusions Interaction data reuse (IDR) is a potential approach for ordinary teachers to developing automated tutoring services n The development of data reuse engine and UI development tools are the next steps n