Rosalind
IBM gave us a challenge to design a concept to use the capabilities of their super computer IBM Watson.
We went into details of IBM Watson and understood its capabilities and features. We decided to use IBM Watson in the domain of education as it is the problem space pretty close to our hearts and we always dream to make education system better. Our challenge was using IBM Watson to help the teacher know if the students are understanding the course being taught or not.
We were motivated to do this project because we were empowering the teachers by letting them know how the course can be improved. Even though this is a small change, it can bring a big change, a better understanding of the course and hence better education for the students taking the course.
The Challenge
The Problem
To find out the capabilities of Watson, we researched where Watson was used previously and we also talked to the people at IBM to know more about Watson. We found our that watson can only parse text data as input which becomes its knowledge bases and can answer questions asked about that data in percentage probability.
We also did primary research by talking to professors and researchers in education and secondary online research to find out about existing technologies used to know student performance and understanding.
We found that knowing where students struggle helps teachers change their course and teaching methods to make it more understandable. Problem is that the existing technologies can only be used for grading and doesn't have any idea of context and trends or patterns in student understanding.
The Solution
We created rosalind, a system which uses IBM Watson to find what parts of the course students understand well and what are the parts student struggle with. Rosalind is not used for grading student work.
The data corpus from which Watson can be trained can only be in text form. In our system Watson can be trained using following data:
•The literature texts related to the course
•Some ideal previously written papers
•Some incorrectly written papers
•Current papers which students write as a part of curriculum
•Any other supplemental sources for grammar and writing style knowledge
Watson can learn about the understanding of the coursework by analysing current papers against the following traits:
•Improper citation
•Being too wordy
•Writing incorrectly
•No right links and connections in what’s being written
•Frequency of citation
•Patterns of mistakes
•Contradictions to the original literature
Once trained, a teacher can use watson to know the understanding of the course content.
Result
With this system, we are empowering teachers with the use of technology by letting them know what part of the course students understand very well and what part they struggle with without spending a lot of time by manually going through student work. Knowing this, the teachers can make changes to their course making their course more understandable and make learning meaningful without spending a lot of time.
My Contribution
I conducted interviews with professors and did secondary research on existing technologies, created personas, did intial concept sketching, concept scenario, and created the final document.