Estimating Answer Strategies using Online Handwritten Data: A Study using Geometry Problems

Emphasis has been placed on logical reasoning skills as a necessary ability and quality for the next generation. Developing logical reasoning skills is one of the objectives of the geometry curriculum in secondary education. Understanding an individual's strategy for finding a solution is necessary to assess his/her logical reasoning abilities. This study is the first to estimate answer strategies using online handwritten data automatically. Using geometry problems as our subject, we: 1) detect symbols written in a diagram; and 2) classify the answer strategies using XGBoost with the features combining uni-grams, bi-grams, and 1-skip-grams of symbols in their written order. Our experimental evaluation with 38 university students for a single geometry problem confirmed that answer strategies were successfully classified into three categories with an accuracy of 0.71, which shows the feasibility of the automatic detection of answer strategies.