Bayes classifier, type tokenizer and NLP module reorganization

This commit is contained in:
mdecimus
2023-10-10 18:58:38 +02:00
parent a0812095ef
commit 3d9efd363a
53 changed files with 4651 additions and 944 deletions

View File

@@ -6,6 +6,7 @@ resolver = "2"
[dependencies]
utils = { path = "../utils" }
nlp = { path = "../nlp" }
maybe-async = { path = "../maybe-async" }
rocksdb = { version = "0.20.1", optional = true }
foundationdb = { version = "0.8.0", features = ["embedded-fdb-include"], optional = true }
@@ -21,13 +22,9 @@ serde = { version = "1.0", features = ["derive"]}
ahash = { version = "0.8.0", features = ["serde"] }
bitpacking = "0.8.4"
lazy_static = "1.4"
whatlang = "0.16" # Language detection
rust-stemmers = "1.2" # Stemmers
tinysegmenter = "0.1" # Japanese tokenizer
jieba-rs = "0.6" # Chinese stemmer
xxhash-rust = { version = "0.8.5", features = ["xxh3"] }
farmhash = "1.1.5"
siphasher = "0.3"
siphasher = "1.0"
parking_lot = "0.12.1"
lru-cache = { version = "0.1.2", optional = true }
num_cpus = { version = "1.15.0", optional = true }