Supplementary materials for AAAI 2022
-
Outcomes_Visual_Inspection_PaperID_10689_Supplementary.pdf: Shows examples of generated information seeking questions (ISQs) by ISEEQ, along with baseline ISQs and ISQs generated using T5 Model fine-tuned on CANARD.
-
Sample_Human_Evaluation_PaperID_10689.pdf: Shows how the human evaluation was performed anonymously. System 1 is the ground truth (test data of QA Discourse, QAMR, and FB Curiosity), System 2 is ISEEQ-ERL (described in the paper), and System 3 is baseline
Other Information:
The file "ISQs_other_QuestionGeneration_types_example.pdf" provides an example scenario that differentiate ISQ from other question generation types in conversational information seeking or conversation AI in general.
For questions or any queries, please feel free to contact: [email protected]