Skip to content

Commit

Permalink
Update readme
Browse files Browse the repository at this point in the history
  • Loading branch information
smilegate-ai committed Oct 11, 2022
1 parent 9abec58 commit 7b2a144
Show file tree
Hide file tree
Showing 2 changed files with 40 additions and 11 deletions.
23 changes: 19 additions & 4 deletions OPELA_Kor_Ver.md
Original file line number Diff line number Diff line change
Expand Up @@ -4,11 +4,18 @@
>이 데이터는 두 역할을 play 하는 대화입니다. 한 사람은 가상의 캐릭터 (또는 "concept")을 가진 페르소나 역할을, 다른 한 사람은 현실에서 마주칠만한 일반적인 사람 역할을 합니다 (이후부터는 편의를 위해 메신저 플랫폼을 이용하는 '사용자'로 호칭합니다).
이는 챗봇과 상호작용하는 상황을 가정하고 이후에 학습할 수 있도록 공감하고 재밌게 반응해주는 챗봇(페르소나)과 사람(사용자)의 관계와 비슷합니다. 두 사람의 대화는 다양한 일상적인 주제에 대해 짧게는 15 turn부터 길게는 80 turn까지 있습니다.

## 본 데이터는 스마일게이트 AI와 서울대학교가 공동 협업으로 진행한 프로젝트입니다.

# 참여연구원
[이윤경](https://yoonkyunglee.oopy.io)<sup>1</sup>, [조원익](https://sites.google.com/site/warnikchow)<sup>2</sup>,배서연<sup>1</sup>, 김지환<sup>2</sup>, 박지상<sup>1</sup>, 김남수<sup>2</sup>, 한소원<sup>1</sup>

<sup>1</sup>[Human Factors Psychology Lab](https://hfpsych.snu.ac.kr/), Seoul National University <br>
<sup>2</sup>[Human Interface Lab](https://hi.snu.ac.kr/), Seoul National University
- [이윤경](https://yoonkyunglee.com)<sup>1</sup>, [조원익](https://sites.google.com/site/warnikchow)<sup>2</sup>,배서연<sup>1</sup>, 김지환<sup>2</sup>, 박지상<sup>1</sup>, 김남수<sup>2</sup>, 한소원<sup>1</sup>,
- 최현우<sup>3</sup>, 황준선<sup>3</sup>, 김무성<sup>3</sup>

<sup>1</sup>[Human Factors Psychology Lab](https://hfpsych.snu.ac.kr/), Department of Psychology, Seoul National University <br>
<sup>2</sup>[Human Interface Lab](https://hi.snu.ac.kr/), Department of Electrical and Computer Engineering, Seoul National University <br>
<sup>3</sup>[Smilegate AI](https://smilegateai.com)



### 데이터 구축 목적
***
Expand Down Expand Up @@ -225,10 +232,19 @@ labelerX에 관한 성분들은 문장 배열에 대한 array로 제공되며,
- 본 데이터에는 사회적으로 이름있는 인물, 브랜드, 혹은 사회적 현상 등에 대한 참가자들의 주관이 반영될 수 있으며, 이는 윤리적으로 문제가 되지 않는 수준이라면 별도로 배제되지 않았습니다.
- 본 데이터에 포함된 설문조사 결과는 대화 당사자들의 의견으로, 제3자의 시선과 다를 수 있습니다.
- 본 데이터에 포함된 태깅 결과에는 각 어노테이터의 주관이 반영될 수 있으며, 이에 따라 문장 별 최종 레이블은 객관적으로 정확한 레이블보다 경향을 파악할 수 있는 가이드 레이블에 가깝습니다.
- 다양한 커뮤니티에서 현실적인 대화를 수집하는 것이 목표였기 때문에 단순재미를 목적으로 유사과학(예: 혈액형 성격, Myers-Briggs Type Indicator 및 기타 유행성 비과학적 신념 등)을 홍보하거나 언급하는 내용은 제거하지 않았습니다. 그러나 앞서 언급한 사례는 과학적 근거가 부족하고 연구의 과학적 목적 및 결과와 관련이 없으니 추후 학술 연구를 위해 이 데이터를 활용하고자 하는 연구자께서는 유의해주시기 바랍니다.


### 인용
```
@misc{SmilegateAI2022OPELA,
title = {OPELA stands for Open-domain conversations by Personas with Empathy, Long-term memory, and Attractive personality.},
author = {Smilegate AI},
year = {2022},
howpublished = {\url{https://github.com/smilegate-ai/OPELA}},
}
```
```
@article{lee2022feels,
title={"Feels like I've known you forever": empathy and self-awareness in human open-domain dialogs},
author={Lee, Yoon Kyung and Cho, Won Ik and Bae, Seoyeon and
Expand All @@ -238,7 +254,6 @@ labelerX에 관한 성분들은 문장 배열에 대한 array로 제공되며,
}
```


### 문의
***
- E -mail : ai_github@smilegate.com
Expand Down
28 changes: 21 additions & 7 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,11 +7,18 @@
This is similar to typical conversation between a chatbot (persona) and a person (user), which assumes a situation of interacting with the chatbot that reacts to the user with empathy and an active attitude.
The conversations range from as short as 15 turns to as long as 80 turns on a variety of everyday topics.

## This data is a joint collaboration project between Smilegate AI and Seoul National University.

# Researchers
[Yoon Kyung Lee](https://yoonkyunglee.oopy.io)<sup>1</sup>, [Won Ik Cho](https://sites.google.com/site/warnikchow)<sup>2</sup>,Seoyeon Bae<sup>1</sup>, Jihwan Kim<sup>2</sup>, Jisang Park<sup>1</sup>, Nam Soo Kim<sup>2</sup>, Sowon Hahn<sup>1</sup>

<sup>1</sup>[Human Factors Psychology Lab](https://hfpsych.snu.ac.kr/), Seoul National University <br>
<sup>2</sup>[Human Interface Lab](https://hi.snu.ac.kr/), Seoul National University
- [Yoon Kyung Lee](https://yoonkyunglee.com)<sup>1</sup>, [Won Ik Cho](https://sites.google.com/site/warnikchow)<sup>2</sup>,Seoyeon Bae<sup>1</sup>, Jihwan Kim<sup>2</sup>, Jisang Park<sup>1</sup>, Nam Soo Kim<sup>2</sup>, Sowon Hahn<sup>1</sup>

- Hyunwoo Choi<sup>3</sup>, Joonsun Hwang<sup>3</sup>, Moosung Kim<sup>3</sup>

<sup>1</sup>[Human Factors Psychology Lab](https://hfpsych.snu.ac.kr/), Department of Psychology, Seoul National University <br>
<sup>2</sup>[Human Interface Lab](https://hi.snu.ac.kr/), Department of Electrical and Computer Engineering, Seoul National University <br>
<sup>3</sup>[Smilegate AI](https://smilegateai.com)


# Purpose of Data Collection
- Many conversation data using personas were presented in a variety of languages, the majority of which were English and Chinese. Persona, on the other hand, was primarily determined by a number of circumstances in specific situations.
Expand Down Expand Up @@ -251,26 +258,33 @@ All attributes are written per statement (per line shift to be exact).

- Because the labeling in this data may reflect the subjectivity of the annotator, it is recommended to utilize these findings as a guide.

- Contents promoting or explaining pseudoscience (determining personality with blood type, the Myers-Briggs Type Indicator, and other popular beliefs, etc.) to the partner for entertainment purposes were not excluded because they were deemed to be a natural conversation. However, there is no scientific support for this information, and it is irrelevant to the study's objective and findings.
- Because our goal was to collect realistic conversation from diverse communities, we didn't remove content advocating or referencing pseudoscience to the conversational partner for enjoyment (e.g., determining personality with blood types, the Myers-Briggs Type Indicator, and other popular non-scientific beliefs, etc.). However, the aforementioned instances lack scientific backing and are unrelated to the study's scientific objectives and findings. Researchers' discretion is advised when utilizing this data for psychology research.



### Cite
```
@misc{SmilegateAI2022OPELA,
title = {OPELA stands for Open-domain conversations by Personas with Empathy, Long-term memory, and Attractive personality.},
author = {Smilegate AI},
year = {2022},
howpublished = {\url{https://github.com/smilegate-ai/OPELA}},
}
```
```
@article{lee2022feels,
title={"Feels like I've known you forever": empathy and self-awareness in human open-domain dialogs},
author={Lee, Yoon Kyung and Cho, Won Ik and Bae, Seoyeon and
Choi, Hyunwoo and Park, Jisang and Kim, Nam Soo and Hahn, Sowon},
year={2022},
publisher={Cognitive Science Society}
publisher={PsyArXiv}
}
```


### Inquiries
***
- E -mail : ai_github@smilegate.com
- Smilegate AI
- 주최 : Smilegate AI 센터
***

![logo_black_gray](https://user-images.githubusercontent.com/95196586/147066863-b9f99434-3ce8-463f-abb4-5e672b3a1fda.png)

0 comments on commit 7b2a144

Please sign in to comment.