Datasets:
File size: 4,131 Bytes
1f4dedc |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 |
"""
Convert Brat format annotations to JSONL format for NER training.
Author: Amir Safari
Date: 17.10.2025
This script processes Brat annotation files (.ann and .txt) from train/dev/test
directories and converts them into JSONL format suitable for NER model training.
"""
import json
import re
from pathlib import Path
print("Starting data conversion from Brat format to JSON Lines...")
# Tag mapping: Create a dictionary to convert tag names to integer IDs
NER_TAGS = [
"O", "B-Taxon", "I-Taxon", "B-Geographical_Location", "I-Geographical_Location",
"B-Habitat", "I-Habitat", "B-Temporal_Expression", "I-Temporal_Expression",
"B-Person", "I-Person",
]
# Create a mapping from tag name to integer ID
tag2id = {tag: i for i, tag in enumerate(NER_TAGS)}
# Process each split directory (train, dev, test)
for split in ["train", "dev", "test"]:
print(f"\nProcessing '{split}' split...")
input_dir = Path(split)
output_file = f"{split}.jsonl"
if not input_dir.exists():
# Skip if directory doesn't exist
print(f"Directory not found: {input_dir}. Skipping split.")
continue
with open(output_file, "w", encoding="utf-8") as outfile:
# Find all .ann files and process them with their corresponding .txt files
ann_files = sorted(input_dir.glob("*.ann"))
for ann_file in ann_files:
txt_file = ann_file.with_suffix(".txt")
if not txt_file.exists():
continue
with open(txt_file, "r", encoding="utf-8") as f:
# Tokenize the text by finding all non-whitespace sequences
text = f.read()
tokens_with_spans = [{"text": match.group(0), "start": match.start(), "end": match.end()} for match in
re.finditer(r'\S+', text)]
if not tokens_with_spans:
continue
tokens = [t["text"] for t in tokens_with_spans]
ner_tags = ["O"] * len(tokens)
# Parse the .ann file to extract entity annotations
with open(ann_file, "r", encoding="utf-8") as f:
annotations = []
# Apply BIO tagging scheme to tokens based on character span overlaps
for line in f:
if not line.startswith("T"): continue
parts = line.strip().split("\t")
if len(parts) < 2: continue
tag_info = parts[1]
tag_parts = tag_info.split(" ")
label = tag_parts[0].replace(" ", "_")
spans_str = " ".join(tag_parts[1:])
char_spans = []
for span_part in spans_str.split(';'):
try:
start, end = map(int, span_part.split(' '))
char_spans.append((start, end))
except ValueError:
continue
if char_spans:
annotations.append({"label": label, "spans": char_spans})
for ann in annotations:
is_first_token = True
for start_char, end_char in ann["spans"]:
for i, token in enumerate(tokens_with_spans):
if token["start"] < end_char and token["end"] > start_char:
ner_tags[i] = f"B-{ann['label']}" if is_first_token else f"I-{ann['label']}"
is_first_token = False
# Convert tag strings to integer IDs for model compatibility
ner_tag_ids = [tag2id.get(tag, tag2id["O"]) for tag in ner_tags]
# Write the processed example as a single JSON line
json_line = json.dumps({
"id": txt_file.stem,
"tokens": tokens,
"ner_tags": ner_tag_ids
})
outfile.write(json_line + "\n")
print(f"Successfully created {output_file}")
print("\nConversion complete! ✨") |