Intelligent Tutoring System for Practical Rhetorical Structure Theory

#Intelligent Tutoring System Development, #Natural Language Processing
#Rhetorical Structure Theory

Duration: 8 weeks, Oct 16th, 2018 - Dec.16th, 2018

Team:Kexin Yang, Pooja Casula, Mingtong Zhang



Rhetorical Structure Theory (RST) is a linguistics framework, frequently used in information retrieval area, that focuses on identifying relations between sentences. The initial reason for building this intelligent tutor is to make learning RST more engaging and efficient.
RST, while being a useful framework that can parse text structure in a genre-free way, has very few learning materials other than academic papers or textbook. I learned this framework by reading academic paper, which was unengaging and time-consuming, then I decided to explore ways to make the RST learning process more effective and interesting.


The strength of this intelligent tutoring system:

  1. It adopts mastery learning and use adaptive problem selection with problems involving unmastered skills, in order to ensure learners reach mastery in each relation.

  2. The problems are presented in sequence from easy to hard, and give contextual feedback to ensure learners stay on the right track.

  3. it is held on a public server, Tutorshop, which enable its free accessibility by anyone interested in learning or teaching RST.

Cognitive Task Analysis (CTA):

We conducted think aloud with four different difficulty factors, on two groups of learners- RST novices and people who have experience with RST (e.g. people who study NLP or linguistics).

We went through three major iterations of the tutoring system.

Low-fi Prototype


Mid-fi Prototype


High-fi Prototype

Photo Photo

Project Report:

Here is the final report for the intelligent tutoring system, which includes our template for CTA (Think Aloud), CTA transcription, Knowledge Component (KC) model and production rules. [Click here for the report]

Try out this Tutoring System:

  1. Go to Tutor shop: [Tutorshop Website]
  2. Choose one from the 10 usernames, and use the password “rst” to log in

    Username: RSTguest1 (The number 1 can be substituted by any number between 1-10)

    Password: rst

  3. Click “Run Problem Set” to start playing with the tutor.

Video demo of Sprinkle