This is a preliminary readme, more information will follow
This repository contains scripts related to the generation of a training database for the Spot Application.
The general idea of the pipeline is the following:
- Bundle similar tags, assign natural language descriptors to create better semantic connections between language and the OSM tagging system
- Generate random artificial queries including area definition, objects (incl. tags and descriptors) and relations/distances
- Call GPT API to generate artificial natural sentences from the draft