Klebert-Engineering / deep-spell-9

Neural Spellcheck, Autocomplete and Fuzzy-match for SQLite FTS5 🤖

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Deep-Spell-9

About

Deep-Spell-9 enables neural auto-completion and matching of geographic queries via deep neural networks.

Figure 1

Web service

A service component is provided which allows you to run Deep-Spell-9 within an interactive browser UI. Use the service.json config file to configure the used models.

You have two options to launch the service:

  1. Via a local Python setup: Clone the repo, and run

    cd [clone-directory]
    pip install -e .
    # Port defaults to 8091
    ./serve.bash [port]
  2. Via Docker:

    docker run --rm -it -p 8091:8091 \
       "ghcr.io/klebert-engineering/ds9:2023.1" \
       "//ds9/serve.bash"

    Note: The default service.json in the image does not provide an FTS database to match a query to ranked results, so you will only be able to see the classifier and completer components in action.

After the service is started, open localhost on the specified port in any browser (localhost:8091) by default.

Web Service With FTS Database Lookup

Download licensed NDS FTS database files according to your coverage needs, and paste them under corpora/. Then launch the Dockerized service as follows:

docker run --rm -it -p 8091:8091 \
   -v ./corpora://ds9/corpora \
   --env SERVICE_CONFIG="corpora/service-with-fts5.json" \
   "ghcr.io/klebert-engineering/ds9:2023.1" \
   "//ds9/serve.bash"

About

Neural Spellcheck, Autocomplete and Fuzzy-match for SQLite FTS5 🤖

License:MIT License


Languages

Language:Python 91.3%Language:HTML 7.5%Language:Shell 1.2%Language:Dockerfile 0.1%