Deconstructing the failure: Analyzing the unanswered questions within educational Q&A

Abstract

Community Question and Answering (CQA) is a well-known platform for knowledge sharing and social learning. CQA services have expanded into the education sector, where school students are the main users. Here, CQA functions as a non-traditional learning environment in which students use their own knowledge to construct the knowledge base. However, there have been cases where some questions are not answered within such a Q&A system. This failure may occur if a question is unclear, complex, inappropriate, or unrelated to the subject in which it is contextualized. While experts do not answer posted questions, co-users moderate responses and help maintain answer quality. However, due to the presence of users from diverse cultural and linguistic backgrounds, many questions remain unanswered. The current study analyzes and explores the failed questions collected from Brainly, a social learning Q&A platform for school students. The quality of 1,000 questions extracted from this service is analyzed based on human-based ratings and extracted textual features. The findings show that a relationship can be drawn between the non-textual assessment results and the objective textually extracted features. This further encourages the study of why a question might be of poor quality. The findings also show which subjects contain the highest number of unanswered questions. These results will further help us to understand how questions should be restructured to obtain answers from their askers' peers.