Abstract
Online discussion has been an important method strongly valued for carrying out collaborative peer learning on the Internet in recent years. However, one of the problems appears to be the quantity of off-topic messages while engaging in the online discussion environment without teacher's intervention. Up to now, several related studies proposing various improved mechanisms or algorithms to detect the occurrence of off-topic messages. Nevertheless, the problem coming before the adoption of any method and in need of being solved is how to determine words or phrases as off-topic content. Therefore, this study covers and analyzes the content based on seven Social Sciences Citation Index (SSCI) papers which explore off-topic messages used by college students who were selected as the subjects for the teaching experiments. The results show that a total of 15229 off-topic messages are found from the students' online discussions. After having a discussion with the Delphi Method, further analysis was made to determine that 260 words or phrases are classified as off-topic messages constantly appearing in online discussions. These words or phrases found in this study can be served as a reference for follow-up research.