Jazz is a JavaScript library that leverages with a Language Learning Model (LLM) backend - GPT 3.5, to interact with data intuitively and effectively. This library streamlines data handling and interaction, making data analysis and data science tasks a breeze.
Jazz empowers users to perform a multitude of operations such as:
- Summarizing data for dataframes and columns.
- Answering questions about the data.
- Conducting descriptive analysis.
- Executing data transformations.
- Creating effective data visualizations.
- Crafting regular expressions (Regex).
- Formatting data using 'Program By Example'.
Jazz was built with the vision to simplify the interaction between data scientists and data analysts with their data. It leverages natural language processing capabilities to create a more intuitive and user-friendly data handling environment.
Jazz works by accepting a user-defined prompt. This prompt is used to select an appropriate tool to handle the task at hand. Available tools are:
- Text to Expressions: Generates regular expressions or date formats.
- Ask: Answer questions about the data
- Code: Performs data cleaning, joining, transformation, and formatting.
- Text: Provides descriptions and insights about the data and columns.
- JSON: Used for data visualization.
These tools return a response that is then sent to the user for further actions.
The user-provided prompts are refined and enhanced using Jazz, which are then forwarded to the LLM. The method used for this enhancement is inspired by the research paper, "Boosting Theory-of-Mind Performance in Large Language Models via Prompting".
The enriched prompt can include a variety of elements such as the dataset, column names, statistics, and more, depending on the selected tool. You can read more about this method here.
Below is a brief overview of how each tool processes the prompts:
- Text to Expressions: User Prompt -> Prompt Enhancement -> LLM -> Generated Expression (Regex, Date Format)
- Ask: User Prompt -> Prompt Enhancement -> LLM -> Python Execution -> Error Handling (If applicable) -> Output Answer
- Code: User Prompt -> Prompt Enhancement -> LLM -> Python Execution -> Error Handling (If applicable) -> Output Code
- Text: User Prompt -> Prompt Enhancement -> LLM -> Output Text
- JSON: User Prompt -> Prompt Enhancement -> LLM -> Output JSON
Welcome to the world of intuitive and natural data interaction with Jazz!