Team Name

query-hub

Team Members

Project Description

Project Description

Finding the right government data is hard—datasets are scattered, inconsistent, and often hidden.

Our Goal: Build a single, user-friendly “national inventory” that aggregates metadata, relationships, and access methods.

Our solution makes data discovery as simple as asking a question: users search in plain language, and an AI-powered engine connects them to the most relevant datasets while revealing unexpected links across sources.

No SQL, pandas, specialized knowledge needed

We’re building a smarter, searchable national data inventory that helps policymakers, researchers, and communities unlock the full value of Australia’s public data.

Key Components

User Interface (UI)

• Accepts inputs through text, voice, and smart uploads.
  
• Supports natural language queries.
  
• Allows users to add additional datasets directly.
  

Processing Layer

• A Large Language Model (LLM) translates user queries into structured searches.
  
• Dynamically builds and executes ABS Data Explorer API calls.
  
• Integrates newly added datasets

Outputs

• Provides relevant data results from government and user-added sources.
  
• Offers multiple visualization options for data exploration and analysis.
  

Overall

Our solution simplifies dataset discovery, integration, and analysis.

By combining natural language querying with a dynamic Data Dictionary and Data Catalog, we empower users to explore, connect, and visualize Australia’s public data—turning a fragmented landscape into a powerful national data inventory.

Data Story

Data Dictionary

Inputs

- Datasets from GovHack datasets

- ABS Data Explorer API & manually downloaded xlxs & csv files
- User-uploaded datasets

Processing

- Parse semi structured data with LLMs to build schema, metadata and understanding
- Resolves inconsistencies and ambiguity across datasets.

Outputs

- An answer to the user's query
- Metadata describing government and user-added datasets.
- Analysis on key trends/insights
- Raw data used to draw conclusions & synthesis and original sources
- Visualizations of raw data that adapt on the fly to returned data

Data Catalog

Inputs

- Datasets from GovHack datasets

- ABS Data Explorer API & csv & xlsx files
- User-uploaded datasets

Processing

- Extracts and stores dataset attributes, including:

- Dataset Name

- Published Date / Collection Date

- Purpose

- Owning Government Department / Data Custodian

- Subject Area(s)

- Abstract / Summary

- Parses structured and semi-structured data.

- Normalize to remove edge cases
- Identifies relationships and dependencies between datasets.

Outputs

- A answer for the user
- Data to back up the answer with sources
- Visualizations to explore the answer more
- A comprehensive inventory of government and additional datasets, including metadata, relationships, and dependencies.