This repository provides the evaluation datasets used in our paper "Comparison of the Intimacy Process between Real and Acting-based Long-term Text Chats" (LREC-COLING2024).
This repository provides two datasets.
We are releasing three weeks' worth of chat logs for four pairs.
label | value |
---|---|
Male pair id | 1, 13 |
Female pair id | 42, 52 |
Num of utterance | 815 |
key | value |
---|---|
room | pair id (unique value for each pair) |
dayid | date number on which the utterance was posted |
speaker | speaker id (unique value for each speaker) |
utterance | message |
We release part of our Japanese Multi-Session Chat. We excluded some pairs for several reasons.
label | value |
---|---|
Num of 5 session pair | 72 |
Num of 3 session pair | 125 |
Num of utterance | 8820 |
key | value |
---|---|
pair_id | persona id |
sid | session id |
tid | turn id |
speaker | speaker id |
utt | message |
sum | persona summary |
interval_sum_label | total interval |
interval_last_label | interval from last session |
key | value |
---|---|
PersonaID | persona id |
A | persona for speaker A |
B | persona for speaker B |
For further details and usage instructions regarding the datasets, please refer to our paper.