uvacw/cart

Conversational Agent Research Toolkit

cart's Issues

Include new assignment method (sequential)

As random_balanced can still lead to issues, sequential can be useful especially for pretests.

        if experimental_design['assignment_method'] == 'sequential':
            sql = 'SELECT conditionid FROM conversations'
            db.commit()
            cur = db.query(sql)
            results = cur.fetchall()
            if len(results) == 0:
                return conditions[0]

            results = [item[0] for item in results]
            last_cond = results[-1]

            total_cond = len(conditions)
            if conditions.index(last_cond) + 1 == total_cond:
                return conditions[0]
            else:
                return conditions[conditions.index(last_cond) + 1]

Change the font/color for redirect to survey

As suggested by one anonymous reviewer:
recommend to change system to automatically advance to survey after chatbot done, or at least put the 'robotic' text in a different color/font, since 'please enter B150 code' puts the chatbot in an experimenter's voice and role, which is not appropriate for a chatbot that could have a different role and can influence people's perception of the chatbot, especially if there's multiple chatbot interactions in a participant's session.