Toward automatic analyses of conversations

Research output: Contribution to conferencePapers


As people can learn and solve problems together that they cannot do alone, automatic analysis of conversations can yield insights for improving them to aid learning and teaching. Such automatic analyses must successfully traverse the obstacle course of voice transcription, categorization, and statistical analysis. Automated transcription is available commercially, and computational linguistics enables automatic categorization. Hence, this talk focuses on automated statistical analysis (statistical discourse analysis [SDA] + artificial intelligence expert system). SDA models (a) pivotal actions that radically change subsequent processes and (b) effects of explanatory variables at multiple levels (sequences of turns/messages, time period, individual, group, organization, etc.) on target actions. The expert system translates a theoretical model into a statistical model, tests it on the data, interprets the results, (if needed, rewrites itself to run a revised analysis), and prints a table of results. Automated SDA is illustrated on 17 students’ 1330 messages in 13 weekly discussions of their lesson designs. Copyright © 2018 CSEPT.
Original languageEnglish
Publication statusPublished - Oct 2018


Chiu, M. M. (2018, October). Toward automatic analyses of conversations. Paper presented at The 13th Cross-Strait Conference on Educational and Psychological Testing Chinese Association of Psychological Testing Annual Conference (CSEPT 2018), Sun Moon Lake Teachers' Hostel‎, Taichung, Taiwan.


Dive into the research topics of 'Toward automatic analyses of conversations'. Together they form a unique fingerprint.