search

Klebert-Engineering / deep-spell-9

Neural Spellcheck, Autocomplete and Fuzzy-match for SQLite FTS5 🤖
MIT License
2 stars 0 forks source link

Add performance-aware serialization/loading of spelling hashes #27

Closed josephbirkner closed 6 years ago

josephbirkner commented 6 years ago

Currently, spelling hashes are stored inefficiently. Also, the k-D tree index has to be rebuilt whenever the service is restarted, which is very slow even on desktop platforms. Therefore, the hashes should be layed out in a binary format that already reflects the lookup index structure.

josephbirkner commented 6 years ago

It should be considered to implement this in SQLite.

josephbirkner commented 6 years ago

On the other hand, low-hanging-fruit may present itself in the form of https://stackoverflow.com/questions/37722991/unable-to-pickle-scipy-spatial-kdtree-objects

josephbirkner commented 6 years ago

ca602cf6814145de15ffafb3a3562f643d186f1a