arxiv-classifier / preprocess_all2023_v2.py
ag2435's picture
added all2023 train-test split with fulltext
2cf1cf5
# ---
# jupyter:
# jupytext:
# cell_metadata_filter: -all
# custom_cell_magics: kql
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.11.2
# kernelspec:
# display_name: arxiv-classifier-next
# language: python
# name: python3
# ---
# %% [markdown]
# Script for converting all 2023 corpus (with full text) to json format.
# %%
#
# parse arguments
#
from argparse import ArgumentParser
parser = ArgumentParser()
# parser.add_argument("--dataset", "-d", type=str, default="minor")
# parser.add_argument("--split", "-s", type=str, default="val")
parser.add_argument("--output_path", "-op", type=str, default="./")
parser.add_argument("--input_path", "-ip", type=str, default="./data",
help="Path to the input data files")
args, opt = parser.parse_known_args()
dataset = 'all2023' # args.dataset
# split = 'val' # args.split
output_path = args.output_path
input_path = args.input_path
# %%
#
# Remaining imports
#
import pandas as pd
import os
import sys
from tqdm import tqdm
from datetime import datetime
from util_preprocess import preprocess_primary_secondary
# %% [markdown]
# Rename columns to reflect standarized terminology:
# - Field: Bio/cs/physics
# - Subfield: Subcategories within each
# - Primary subfield (bio., cs.LG): Given primary subfield, you can infer the field
# - Secondary subfields: Includes primary subfield, but also includes any subfields that were tagged in the paper (1-5)
#
# Old terminology to standardized terminology translation:
# - Prime category = Primary subfield
# - Abstract category = secondary subfield
# - Major category = field
# %%
file_path = os.path.join(
input_path,
'raw',
'all2023',
'all-2023-v2.tsv'
)
save_dir = os.path.join(output_path, dataset)
os.makedirs(save_dir, exist_ok=True)
train_save_path = os.path.join(save_dir, "train.json")
test_save_path = os.path.join(save_dir, "test.json")
print("Reading from file: ", file_path)
FEATURES = [
'paper_id',
'version',
'yymm',
'created',
'title',
'secondary_subfield',
'abstract',
'primary_subfield',
'field',
'fulltext'
]
def get_date(paper):
# first get the arxiv id
paper_id = paper['paper_id']
# yy is the first two characters of the arxiv id
yy = paper_id[:2]
# mm is the next two characters of the arxiv id
mm = paper_id[2:4]
# return datatime object
date_obj = datetime(year=int(f"20{yy}"), month=int(mm), day=1)
# return date as string in the format YYYY-MM-DDTHH:MM:SS
return date_obj #.strftime('%Y-%m-%dT%H:%M:%S')
def get_fulltext(row):
filename = f"{row['paper_id']}v{row['version']}.txt"
filepath = os.path.join(
input_path,
'raw',
'all2023',
'fulltext',
row['paper_id'].split('.')[0], # get yymm subfolder
filename,
)
try:
with open(filepath, 'r') as f:
fulltext = f.read()
except FileNotFoundError:
fulltext = None
return fulltext
count_orig = []
count_isvalid = []
count_hasfulltext = []
# Ignore all quotes by setting quoting=3 (corresponds to csv.QUOTE_NONE)
for i, original_df in tqdm(enumerate(pd.read_csv(file_path, sep='\t', chunksize=10, dtype=str, quoting=3))):
count_orig.append( len(original_df) )
# set `yymm` column
original_df.loc[:, 'yymm'] = original_df['paper_id'].apply(lambda x: x[:4])
# set `created` column using yymm from the paper_id as a proxy
original_df.loc[:, 'created'] = original_df.apply(lambda x: get_date(x), axis=1)
# set non-existent features to empty string
for feature in FEATURES:
if feature not in original_df.columns:
original_df[feature] = ""
# convert categories to primary and secondary subfield
original_df.loc[:, 'primary_subfield'] = original_df['categories'].apply(lambda x : x.split()[0])
# keep the secondary_subfield a string, will be converted to list in preprocess_primary_secondary()
original_df.loc[:, 'secondary_subfield'] = original_df['categories']
# Apply preprocessing rules and add `field` column
df = preprocess_primary_secondary(original_df)
count_isvalid.append( len(df) )
# Get fulltext for each paper from .txt files
df.loc[:, 'fulltext'] = df.apply(get_fulltext, axis=1)
# drop `categories` column
df = df.drop(columns=['categories'])
# print number of rows where fulltext is not None
df = df.dropna(subset=['fulltext'])
count_hasfulltext.append( len(df) )
# assert that df has all the required columns
assert all([feature in df.columns for feature in FEATURES]), f"Missing features in df: {FEATURES}"
# convert all columns to string per huggingface requirements
DATE_CUTOFF = datetime(2023, 7, 1)
if i == 0:
# save the rows with created date before 2023-06-01 to train.json
df[df['created'] < DATE_CUTOFF].to_json(train_save_path, lines=True, orient="records") #, date_format='iso')
# save the rows with created date after 2023-06-01 to test.json
df[df['created'] >= DATE_CUTOFF].to_json(test_save_path, lines=True, orient="records") #, date_format='iso')
else:
# save the rows with created date before 2023-06-01 to train.json
df[df['created'] < DATE_CUTOFF].to_json(train_save_path, lines=True, orient="records", mode='a')#, date_format='iso')
# save the rows with created date after 2023-06-01 to test.json
df[df['created'] >= DATE_CUTOFF].to_json(test_save_path, lines=True, orient="records", mode='a')#, date_format='iso')
# print("Saved to: ", save_path)
df_err = pd.DataFrame({
'orig': count_orig,
'isvalid': count_isvalid,
'hasfulltext': count_hasfulltext,
})
df_err.to_csv(f"error_log.csv", index=False)
# %%