asu-kim/conversation-calendar

---
license: bsd-3-clause
task_categories:
- question-answering
language:
- en
pretty_name: Conversation with Calendar Items Dataset
size_categories:
- 1K<n<10K
---
# Overview
This dataset was generated for evaluation purposes, focusing on compiling an individual’s information with questions and answers.
- **Use of ChatGPT-4o**: Used for generating human-like text and structured calendar data.
- **Format**: For easy processing, data is stored in JSON (calendar data) and TXT (conversation data) formats.
- **Persona**: The dataset is centered around an individual named ‘Alex’.
- **Event Generation**: Calendar events are designed to be realistic rather than randomly generated through ChatGPT-4o with prompt engineering.
- **Conversational Data**: Relevant to calendar events so that the conversation context is natural for Alex, Alex's family, and Alex's friends.
The dataset includes seven conversational topics based on day-to-day discussions.

# Addtional Information
This dataset has been released as part of our research work, **personal-llm-kg**, which is available in different formats below.
- [The link to our source code on GitHub](https://github.com/asu-kim/personal-llm-kg.git)
- [The link to our video](https://www.youtube.com/watch?v=lwW8FWrzwzM)
- [The link to our paper published at the WWW'25 conference](https://doi.org/10.1145/3701716.3715473) - Title: "Personalization of Large Language Models using Retrieval Augmented Generation and Knowledge Graphs"

# Authors
Deeksha Prahlad ([e-mail]) and
Hokeun Kim ([e-mail])