Towards a conversational agent for story reading

• Main Author: Chan, Lynette Danielle Ko
• Format: text
• Language: English
• Published: Animo Repository 2018
• Subjects: Reading; Reading comprehension; Storytelling in education
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/5529
• Institution: De La Salle University

Students assigned to a reading task often progress very slowly or even abandon the task entirely due to certain factors, such as the lack of interest which causes boredom, poor reading comprehension which causes discouragement, and disengagement from the reading material which can ultimately cause the reader to stop from reading. Various studies have reported that these negative academic moods can be mediated through having meaningful conversation with the children during reading. While interventions such as collaborative reading have been proposed to foster reading motivation in the classroom and conversational agents have been shown to be just as effective as humans in engaging students, very few studies have considered the application of conversational agents as storytelling peers towards encouraging and scaffolding the learning of students by collaborative learning through discussion. In this research, a conversational agent that plans various dialogue moves to address the causes of reading problems through story content elaboration and conversation was developed. The agent derives its knowledge from a computational model of a story and a lexical database of word meanings derived from WordNet. Results from the evaluation with children who are between 10 to 16 years old show that the interest of the users in conversing with the agent and the engagement of the users in the reading task affect how they will react to the agent's questions and assertions. Furthermore, rather than supplying answers to user questions directly, collaborating with the user to arrive at the answer through discussion proves more effective in addressing the reading problems of children.