conv_ai_2 / README.md
lhoestq's picture
lhoestq HF staff
add dataset_info in dataset metadata
47919ed
metadata
annotations_creators:
  - crowdsourced
language_creators:
  - crowdsourced
language:
  - en
license:
  - unknown
multilinguality:
  - monolingual
size_categories:
  - 1K<n<10K
source_datasets:
  - original
task_categories:
  - conversational
  - text-classification
task_ids:
  - text-scoring
paperswithcode_id: convai2
pretty_name: Conversational Intelligence Challenge 2
tags:
  - evaluating-dialogue-systems
dataset_info:
  features:
    - name: id
      dtype: string
    - name: dialog_id
      dtype: string
    - name: dialog
      list:
        - name: id
          dtype: int32
        - name: sender
          dtype: string
        - name: text
          dtype: string
        - name: sender_class
          dtype: string
    - name: bot_profile
      sequence:
        list: string
    - name: user_profile
      sequence:
        list: string
    - name: eval_score
      dtype: int32
    - name: profile_match
      dtype: int32
  config_name: conv_ai_2
  splits:
    - name: train
      num_bytes: 8403805
      num_examples: 3495
  download_size: 6636788
  dataset_size: 8403805

Dataset Card for conv_ai_2

Table of Contents

Dataset Description

Dataset Summary

ConvAI is a dataset of human-to-bot conversations labeled for quality. This data can be used to train a metric for evaluating dialogue systems. Moreover, it can be used in the development of chatbots themselves: it contains information on the quality of utterances and entire dialogues, that can guide a dialogue system in search of better answers.

Supported Tasks and Leaderboards

[More Information Needed]

Languages

[More Information Needed]

Dataset Structure

Data Instances

{
        "dialog_id": "0x648cc5b7",
        "dialog": [
            {
                "id": 0,
                "sender": "participant2",
                "text": "Hi! How is your day? \ud83d\ude09",
                "sender_class": "Bot"
            },
            {
                "id": 1,
                "sender": "participant1",
                "text": "Hi! Great!",
                "sender_class": "Human"
            },
            {
                "id": 2,
                "sender": "participant2",
                "text": "I am good thanks for asking are you currently in high school?",
                "sender_class": "Bot"
            }
        ],
        "bot_profile": [
            "my current goal is to run a k.",
            "when i grow up i want to be a physical therapist.",
            "i'm currently in high school.",
            "i make straight as in school.",
            "i won homecoming queen this year."
        ],
        "user_profile": [
            "my favorite color is red.",
            "i enjoy listening to classical music.",
            "i'm a christian.",
            "i can drive a tractor."
        ],
        "eval_score": 4,
        "profile_match": 1
    }

Data Fields

  • dialog_id : specifies the unique ID for the dialogs.
  • dialog : Array of dialogs.
  • bot_profile : Bot annotated response that will be used for evaluation.
  • user_profile : user annoted response that will be used for evaluation.
  • eval_score : (1, 2, 3, 4, 5) how does an user like a conversation. The missing values are replaced with -1
  • profile_match : (0, 1) an user is given by two profile descriptions (4 sentences each), one of them is the one given to the bot it had been talking to, the other one is random; the user needs to choose one of them.The missing values are replaced with -1

Data Splits

[More Information Needed]

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

@article{DBLP:journals/corr/abs-1902-00098, author = {Emily Dinan and Varvara Logacheva and Valentin Malykh and Alexander H. Miller and Kurt Shuster and Jack Urbanek and Douwe Kiela and Arthur Szlam and Iulian Serban and Ryan Lowe and Shrimai Prabhumoye and Alan W. Black and Alexander I. Rudnicky and Jason Williams and Joelle Pineau and Mikhail S. Burtsev and Jason Weston}, title = {The Second Conversational Intelligence Challenge (ConvAI2)}, journal = {CoRR}, volume = {abs/1902.00098}, year = {2019}, url = {http://arxiv.org/abs/1902.00098}, archivePrefix = {arXiv}, eprint = {1902.00098}, timestamp = {Wed, 07 Oct 2020 11:09:41 +0200}, biburl = {https://dblp.org/rec/journals/corr/abs-1902-00098.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} }

Contributions

Thanks to @rkc007 for adding this dataset.