jameynakama/randsense-api

Randsense

    verb_types = [
        "intran",
        "tran",
        "ditran",
        "link"
    ]

    pronoun_types = [
        "dem",
        "obj",
        "subj",
        "poss",  # their
        "possnom",  # theirs
        "refl",
        "poss",
        "univ",
        "indef(assert)",
        "indef(neg)",
        "indef(nonassert)",
    ]

Usage

You must have docker installed to run this app (the easy way).

You also must have a lexicon to ingest. I don't include it in the repo because it's very large, so please just ask me for it if you'd like to play with the app yourself.

First time:

$ docker-compose build --pull
$ docker-compose run --rm manage ingest_lexicon LEXICON.xml
# Run this next command for each word class.
# Words with a rank of 0 will not be returned by queries.
$ docker-compose run --rm manage rank_words_by_frequency --word-class=noun unigram_freq.csv
$ docker-compose run --rm manage rank_words_by_frequency --word-class=verb unigram_freq.csv
$ docker-compose run --rm manage rank_words_by_frequency --word-class=adjective unigram_freq.csv
# Etc. etc.

General running:

$ docker-compose up web

Maintenance

When adding a new class, ensure it gets put into the the models map and any management commands.

Whenever changing how ingestion works, you need to re-ingest and create new test fixtures.