metadata

annotations_creators:
  - machine-generated
language:
  - en
language_creators:
  - found
license:
  - unknown
multilinguality:
  - monolingual
pretty_name: CCI Dataset V2
size_categories:
  - 10K<n<100K
source_datasets:
  - original
task_categories:
  - text-classification
  - feature-extraction
task_ids:
  - intent-classification
  - sentiment-classification
  - topic-classification
  - multi-class-classification
  - sentiment-analysis
paperswithcode_id: null

CCI Dataset V2

Dataset Description

Repository: raghavdw/cci-dataset-v2
Point of Contact: [Please add your contact information]

Dataset Summary

The CCI (Customer Conversation Intelligence) Dataset V2 is a comprehensive collection of airline customer service interactions from Intelligent Virtual Assistant (IVA) conversations. The dataset contains 15,598 entries with rich annotations including intent prediction, sentiment analysis, empathy scoring, and conversation topic classification. Originally sourced from airline customer service interactions, this dataset provides valuable insights into how virtual assistants handle customer inquiries in the aviation sector.

Supported Tasks and Leaderboards

Conversation Intent Classification: Predict the intent behind customer utterances
Sentiment Analysis: Analyze the sentiment of customer messages
Empathy Detection: Score conversations based on demonstrated empathy
Active Listening Assessment: Evaluate the quality of listening in conversations
Topic Classification: Categorize conversations into specific topics
RSIC (Response Style Intelligence Classification): Analyze and score conversation response styles

Languages

The dataset primarily contains conversations in English (to be confirmed).

Dataset Structure

The dataset is split into training (80%) and testing (20%) sets.

Data Instances

Each instance in the dataset represents a single utterance or conversation turn with the following features:

{
    'Index': int,  # Unique identifier
    'Utterance': str,  # The actual conversation text
    'Predicted_Intent': str,  # Classified intent of the utterance
    'Intent_Score': float,  # Confidence score of intent prediction
    'Sentiment': str,  # Analyzed sentiment
    'empathy_score': int,  # Numerical score for empathy
    'listening_score': int,  # Numerical score for active listening
    'Topic': int,  # Numerical topic identifier
    'Topic_Name': str,  # Name/description of the topic
    'rsic_score': float,  # Response Style Intelligence Classification score
    'RSICs': str,  # Response Style Intelligence Classification categories
    'fallback_type': str  # Type of fallback response if applicable
}

Data Fields

Field Name	Type	Description
Index	Integer	Unique identifier for each entry
Utterance	String	The actual text of the conversation turn
Predicted_Intent	String	The predicted intent of the utterance
Intent_Score	Float	Confidence score for the intent prediction (0-1)
Sentiment	String	Sentiment classification of the utterance
empathy_score	Integer	Numerical score indicating level of empathy
listening_score	Integer	Numerical score indicating active listening quality
Topic	Integer	Numerical identifier for conversation topic
Topic_Name	String	Descriptive name of the conversation topic
rsic_score	Float	Response Style Intelligence Classification score
RSICs	String	RSIC categories/labels
fallback_type	String	Type of fallback response if applicable

Data Splits

The dataset is divided into:

Training set: 12,478 examples (80%)
Test set: 3,120 examples (20%)

Dataset Creation

Curation Rationale

This dataset is created as part of a capstone project for a certification program in MLOps that is building a customer intelligence platform that can be used customer service chatbots, BERT classifiers and analytics and insights platforms for ML training and customer service enhancements

Source Data

This dataset is derived from the "Relational Strategies in Customer Service (RSICS)" dataset originally published on Kaggle (https://www.kaggle.com/datasets/veeralakrishna/relational-strategies-in-customer-servicersics). The source data consists of airline IVA (Intelligent Virtual Assistant) conversations, providing a real-world context for customer service interactions in the aviation industry.

The original dataset was specifically designed to study relational strategies in customer service contexts, with a focus on how virtual assistants handle various types of customer inquiries and situations in the airline industry. This makes it particularly valuable for developing and improving customer service AI systems in high-stakes service environments.

Annotations

Annotations are spot checked by human evaluators for sentiment and predicted_intent primarily

Considerations for Using the Data

Social Impact of Dataset

This dataset can be used to improve customer service interactions by:

Enhancing empathy in automated responses
Improving intent recognition accuracy
Developing better conversation flow understanding
Training more effective customer service systems

Discussion of Biases

Other Known Limitations

The dataset may contain specific patterns or biases based on the source of customer service interactions
Intent and sentiment predictions are model-generated and may contain inherent biases or errors

Additional Information

Dataset Curators

[To be added: Information about the team or individuals who created and maintain this dataset]

Licensing Information

[To be added: License information for the dataset]

Citation Information

[To be added: How to cite this dataset in academic work]

Contributions

Thanks to @raghavdw for adding this dataset.