Downloads: 1
Diff of
/src/constants/n2c2.py
[000000]
..
[735bb5]
Switch to unified view
| a | b/src/constants/n2c2.py | ||
|---|---|---|---|
| 1 | # coding: utf-8 |
||
| 2 | |||
| 3 | # Base Dependencies |
||
| 4 | # ----------------- |
||
| 5 | import re |
||
| 6 | from pathlib import Path |
||
| 7 | from os.path import join as pjoin |
||
| 8 | |||
| 9 | |||
| 10 | N2C2_VOCAB_PATH = Path(pjoin("data", "n2c2", "vocab", "vocab.txt")) |
||
| 11 | |||
| 12 | N2C2_PATH = Path("data/n2c2") |
||
| 13 | N2C2_ENTITY_TYPES = [ |
||
| 14 | "Drug", |
||
| 15 | "Strength", |
||
| 16 | "Duration", |
||
| 17 | "Route", |
||
| 18 | "Form", |
||
| 19 | "ADE", |
||
| 20 | "Dosage", |
||
| 21 | "Reason", |
||
| 22 | "Frequency", |
||
| 23 | ] |
||
| 24 | N2C2_ATTR_TYPES = [ |
||
| 25 | "Strength", |
||
| 26 | "Duration", |
||
| 27 | "Route", |
||
| 28 | "Form", |
||
| 29 | "ADE", |
||
| 30 | "Dosage", |
||
| 31 | "Reason", |
||
| 32 | "Frequency", |
||
| 33 | ] |
||
| 34 | |||
| 35 | N2C2_REL_TYPES = [ |
||
| 36 | "Strength-Drug", |
||
| 37 | "Duration-Drug", |
||
| 38 | "Route-Drug", |
||
| 39 | "Form-Drug", |
||
| 40 | "ADE-Drug", |
||
| 41 | "Dosage-Drug", |
||
| 42 | "Reason-Drug", |
||
| 43 | "Frequency-Drug", |
||
| 44 | ] |
||
| 45 | |||
| 46 | N2C2_REL_TEST_WEIGHTS = [ |
||
| 47 | 10255 / 41086, |
||
| 48 | 568 / 41086, |
||
| 49 | 6784 / 41086, |
||
| 50 | 5382 / 41086, |
||
| 51 | 981 / 41086, |
||
| 52 | 3563 / 41086, |
||
| 53 | 4335 / 41086, |
||
| 54 | 9218 / 41086, |
||
| 55 | ] |
||
| 56 | |||
| 57 | N2C2_ATTR_ENTITY_CANDIDATES = { |
||
| 58 | "Strength": ["Drug"], |
||
| 59 | "Duration": ["Drug"], |
||
| 60 | "Route": ["Drug"], |
||
| 61 | "Form": ["Drug"], |
||
| 62 | "ADE": ["Drug"], |
||
| 63 | "Dosage": ["Drug"], |
||
| 64 | "Reason": ["Drug"], |
||
| 65 | "Frequency": ["Drug"], |
||
| 66 | } |
||
| 67 | |||
| 68 | N2C2_SPLITS_DIR = "data/n2c2/splits" |
||
| 69 | N2C2_HF_TRAIN_PATH = "data/n2c2/train.hf" |
||
| 70 | N2C2_HF_TEST_PATH = "data/n2c2/test.hf" |
||
| 71 | |||
| 72 | |||
| 73 | N2C2_ANNONYM_PATTERNS = { |
||
| 74 | "hour": re.compile(r"\[\*\*\d+-\d+\*\*\]\s*PM"), |
||
| 75 | "date": re.compile( |
||
| 76 | r"(\[\*\*(Date|Month|Year)[^\*]*\*\*\])|(\[\*\*\d+-\d+-?\d*\*\*\])" |
||
| 77 | ), |
||
| 78 | "hospital": re.compile(r"(\[\*\*[^\*]*Hospital[^\*]*\*\*\])"), |
||
| 79 | "name": re.compile(r"(\[\*\*[^\*]*(Name|name)[^\*]*\*\*\])"), |
||
| 80 | "telephone": re.compile(r"(\[\*\*[^\*]*(Telephone|telephone)[^\*]*\*\*\])"), |
||
| 81 | "location": re.compile(r"(\[\*\*[^\*]*(Location|location|\d+-/\d+)[^\*]*\*\*\])"), |
||
| 82 | "address": re.compile(r"(\[\*\*[^\*]*(Address|address|Country|State)[^\*]*\*\*\])"), |
||
| 83 | "age": re.compile(r"(\[\*\*[^\*]*(Age)[^\*]*\*\*\])"), |
||
| 84 | "number": re.compile( |
||
| 85 | r"(\[\*\*[^\*]*(Number|Numeric Identifier|number)[^\*]*\*\*\])|(\[\*\*\d+\*\*\])" |
||
| 86 | ), |
||
| 87 | } |
||
| 88 | |||
| 89 | N2C2_IOB_TAGS = [ |
||
| 90 | "O", |
||
| 91 | "B-Drug", |
||
| 92 | "I-Drug", |
||
| 93 | "B-Strength", |
||
| 94 | "I-Strength", |
||
| 95 | "B-Duration", |
||
| 96 | "I-Duration", |
||
| 97 | "B-Route", |
||
| 98 | "I-Route", |
||
| 99 | "B-Form", |
||
| 100 | "I-Form", |
||
| 101 | "B-ADE", |
||
| 102 | "I-ADE", |
||
| 103 | "B-Dosage", |
||
| 104 | "I-Dosage", |
||
| 105 | "B-Reason", |
||
| 106 | "I-Reason", |
||
| 107 | "B-Frequency", |
||
| 108 | "I-Frequency", |
||
| 109 | ] |
||
| 110 | |||
| 111 | N2C2_RD_MAX = 30 |