Data for Multi-agent learning in multi-domain spoken dialogue systems
This dataset correspond to the results presented in NIPS 2015 Workshop on Spoken Language Understanding and Interaction, paper MULTI-AGENT LEARNING IN MULTI-DOMAIN SPOKEN DIALOGUE SYSTEMS and relates to Fig. 1 Two contrasts were presented COMMSCALE and MBCM. MBCM_L6_[1,2,3] is previously obtained data https://www.repository.cam.ac.uk/handle/1810/251169 . COMMSCALE_L6_[1,2,3] is the data obtained in interaction with Amazon MTurk while training three policies for L6 domain using multi-agent learning. In each directory there is a call directory with a time stamp in the name which contains session.xml file with the dialogue log and feedback.xml file with the user feedback.