README.md

metadata

license: apache-2.0
language:
  - en
tags:
  - dialogue segmentation
size_categories:
  - 1K<n<10K

Dataset Card for SuperDialseg

Table of Contents

• Table of Contents
• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Source Data
  • Annotations
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
  • Other Known Limitations
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information
  • Contributions

Dataset Description

• Homepage:
• Repository:
• Paper: SuperDialseg: A Large-scale Dataset for Supervised Dialogue Segmentation
• Leaderboard: https://github.com/Coldog2333/SuperDialseg
• Point of Contact: jiangjf@is.s.u-tokyo.ac.jp

Dataset Summary

[More Information Needed]

Supported Tasks and Leaderboards

[More Information Needed]

Languages: English

Dataset Structure

Data Instances

{
  "dial_data": {
    "super_dialseg": [
      {
        "dial_id": "8df07b7a98990db27c395cb1f68a962e",
        "turns": [
          {
            "da": "query_condition",
            "role": "user",
            "turn_id": 0,
            "utterance": "Hello, I forgot o update my address, can you help me with that?",
            "topic_id": 0,
            "segmentation_label": 0
          },
          ...
          {
            "da": "respond_solution",
            "role": "agent",
            "turn_id": 11,
            "utterance": "DO NOT contact the New York State DMV to dispute whether you violated a toll regulation or failed to pay the toll , fees or other charges",
            "topic_id": 4,
            "segmentation_label": 0
          }
        ],
      ...
      }
    ]
}

Data Fields

Dialogue-Level

• dial_id: ID of a dialogue;
• turns: All utterances of a dialogue.

Utterance-Level

• da: Dialogue Act annotation derived from the original DGDS dataset;
• role: Role annotation derived from the original DGDS dataset;
• turn_id: ID of an utterance;
• utterance: Text of the utterance;
• topic_id: ID (order) of the current topic;
• segmentation_label: 1: it is the end of a topic; 0: others.

Data Splits

SuperDialseg follows the dataset splits of the original DGDS dataset.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

SuperDialseg was built on the top of doc2dial and MultiDoc2dial datasets. Please refer to the original papers for more details.

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

The annotation of dialogue segmentation points is constructed by a set of well-designed strategy. Please refer to the paper for more details.

Other annotations like Dialogue Act and Role information are derived from doc2dial and MultiDoc2dial datasets.

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

Apache License Version 2.0, following the licenses of doc2dial and MultiDoc2dial.

Citation Information

Coming soon

Contributions

Thanks to @Coldog2333 for adding this dataset.