Automated topic spotting provides added efficiency in a chat based tutoring environment

Abstract

Dr Math is a mobile, online tutoring project which allows primary and secondary school pupils to contact tutors in mathematics using text based chat systems on their cell phones. The tutors use traditional Internet based workstations. Dr Math is extremely popular with school pupils and Dr Math tutors are often chatting to twenty pupils concurrently. It is important that the tutors are as efficient as possible. This paper describes the implementation of an automated topic spotter in a mobile chat-based tutoring environment. The tutors, although well versed in mathematics, were not all home language English speakers. They often struggled with the cryptic abbreviations which the pupils used in their queries. In addition, the number of pupils asking questions far exceeded the number of tutors available. It became important to ensure that the tutors were as efficient as possible. The automated topic spotter would process the cryptic questions, determine the general mathematical topic, and automatically provide supporting material to the tutors. The learning gained on this project could be easily reimplemented in other domains besides education including health and public services.