I’m developing a geospatial data viewer that connects to various services (WMS, WMTS). I have a server that clients connect to. When a client makes a request, I cache it on the server. However, I would like to start prefetching likely to be requested by the client tiles. In researching this I’ve come across many prefetching algorithms in research papers: Retrospective Adaptive Prefetch is a user driven algorithm that applies heuristics to recent user actions to determine what tiles need to be prefetched. Tile Prefetching Based on Previous k Movements prefetches tiles based on recent user movement. Distributed Caching System Based on Spatiotemporal applies a heuristic based on when user movements occurred.
The issue I’m running into is these all these algorithms require user interaction data to get started.
Is there a source that has user interaction data with GIS sources?
I know places like OpenStreetMap have data that shows the total amount each tile was accessed in a day, but they don’t share the user trace of accessing those tiles.
I’m working in Python and it’s possible I’ve missed some great package that does geospatial prefetching tile prediction. I seek any easy solutions like this.