removed restriction on paper versions for major, minor, all2023

README.md

@@ -35,6 +35,16 @@ Original data: https://www.dropbox.com/scl/fo/wwu0ifghw4sco09g67frb/h?rlkey=6ddg
 
 <!-- Working doc: https://docs.google.com/document/d/1ZTiR5M7NBpsfYbJdlP6qcIldBJVzJI1Hm8srgI8hfnI/edit?usp=sharing -->
 
+## Set up dependencies
+
+Get subfields to ignore and subfield aliases:
+```
+# in your conda env
+git clone https://github.com/ag2435/arxiv-classifier-next
+cd arxiv-classifier-next
+conda develop .
+```
+
 ## Preprocessing
 
 Transform raw data into JSON format:

all2023/val.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e6c4607bdb7d64e7a7e39d24fe2585d45c392d80707f3b8cb935fee68fe524f
-size 285350475
+oid sha256:728f5ae7ff4f3907c9a5e6f3b35d1a2968a02cd825f50dc575efe94c7d4e4ffa
+size 409838242

major/test.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f479ccd1901ebd967a64a17b6ec9934501629f5caa7bc88b9fb8ca512f54011a
-size 216694822
+oid sha256:c7849149e5f504fd8f99efa3b5f1c2d65b541b2b76640d563bababa4b4f6d395
+size 406464012

major/train.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:258b7304c47a12f0c035657831926953207399fccbcfa9018817b514a2cbff7b
-size 863033790
+oid sha256:37aa1843751273b4e27633b1ec1a0fad6256a9143ed48d81bbf34ffae03116c4
+size 1639035113

minor/test.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fb4a6091612f356e4c6184bae2468d4ac5f5fc41729d947b1eaec82017b06ed5
-size 860652299
+oid sha256:667308dc5ddfa3cf17869673427e5ce85abf054fee301e8a1cf58ee04e6d9d3a
+size 1453222079

minor/train.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93e6352bf9f03225a256f0e44a83b4de3b608b4024ac78ba3b808e9b0db223b3
-size 3410271223
+oid sha256:8834e6cfa89d589cffcc5335a3bc2371ce91f6627c4a653ec689b3e370582b8e
+size 5823161950

preprocess_all2023.py

@@ -37,23 +37,9 @@ import pandas as pd
 import numpy as np
 import os
 from tqdm import tqdm
-
-# %%
-IGNORED_CATEGOREIES = set([
-    "atom-ph", "bayes-an", "chao-dyn", "chem-ph", "cmp-lg", "comp-gas", "cond-mat", "dg-ga", "funct-an",
-    "mtrl-th", "patt-sol", "plasm-ph", "q-alg", "q-bio", "solv-int", "supr-con", "test", "test.dis-nn", 
-    "test.mes-hall", "test.mtrl-sci", "test.soft", "test.stat-mech", "test.str-el", "test.supr-con"
-])
-print("number of ignored categories: ", len(IGNORED_CATEGOREIES))
-CATEGORY_ALIASES = {
-    'math.MP': 'math-ph',
-    'stat.TH': 'math.ST',
-    'math.IT': 'cs.IT',
-    'q-fin.EC': 'econ.GN',
-    'cs.SY': 'eess.SY',
-    'cs.NA': 'math.NA'
-}
-print("number of category aliases: ", len(CATEGORY_ALIASES))
+from arxiv_classifier.utils import (CATEGORY_ALIASES, 
+                                    GENERAL_CATEGORIES, 
+                                    IGNORED_CATEGOREIES)
 
 # %% [markdown]
 # Rename columns to reflect standarized terminology:
@@ -104,16 +90,15 @@ for i, original_df in tqdm(enumerate(pd.read_csv(file_path, sep='\t', chunksize=
             original_df[feature] = ""
 
     # ignore entries from ignored categories (i.e., primary subfield)
-    df = original_df[~original_df['primary_subfield'].isin(IGNORED_CATEGOREIES)]
-    df = df[(df['version'] == 1)]
+    df = original_df[~original_df['primary_subfield'].isin(GENERAL_CATEGORIES | IGNORED_CATEGOREIES)]
     if df['primary_subfield'].isin(CATEGORY_ALIASES.keys()).sum():
         ValueError("IMPLEMENT CODE TO CHANGE CATEGORY_ALIASES ")
 
     # convert all columns to string per huggingface requirements
     if i == 0:
-        df.astype(str).to_json(save_path, lines=True, orient="records")
+        df.to_json(save_path, lines=True, orient="records")
     else:
-        df.astype(str).to_json(save_path, lines=True, orient="records", mode='a')
+        df.to_json(save_path, lines=True, orient="records", mode='a')
 
 print("Saved to: ", save_path)
 # %%

preprocess_major_minor.py

@@ -37,23 +37,9 @@ import pandas as pd
 import numpy as np
 import os
 from tqdm import tqdm
-
-# %%
-IGNORED_CATEGOREIES = set([
-    "atom-ph", "bayes-an", "chao-dyn", "chem-ph", "cmp-lg", "comp-gas", "cond-mat", "dg-ga", "funct-an",
-    "mtrl-th", "patt-sol", "plasm-ph", "q-alg", "q-bio", "solv-int", "supr-con", "test", "test.dis-nn", 
-    "test.mes-hall", "test.mtrl-sci", "test.soft", "test.stat-mech", "test.str-el", "test.supr-con"
-])
-print("number of ignored categories: ", len(IGNORED_CATEGOREIES))
-CATEGORY_ALIASES = {
-    'math.MP': 'math-ph',
-    'stat.TH': 'math.ST',
-    'math.IT': 'cs.IT',
-    'q-fin.EC': 'econ.GN',
-    'cs.SY': 'eess.SY',
-    'cs.NA': 'math.NA'
-}
-print("number of category aliases: ", len(CATEGORY_ALIASES))
+from arxiv_classifier.utils import (CATEGORY_ALIASES, 
+                                    GENERAL_CATEGORIES, 
+                                    IGNORED_CATEGOREIES)
 
 # %% [markdown]
 # Rename columns to reflect standarized terminology:
@@ -74,16 +60,7 @@ os.makedirs(save_dir, exist_ok=True)
 save_path = os.path.join(save_dir, f"{split}.json")
 print("Reading from file: ", file_path)
 
-# # delete file if it exists
-# if os.path.exists(save_path):
-#     print("Overwriting existing file: ", save_path)
-#     os.remove(save_path)
-
-# with open(save_path, 'a') as f:
-for i, original_df in tqdm(enumerate(pd.read_json(file_path, lines=True, chunksize=10**4))):
-    # print(f"Chunk {i+1}")
-    # print("original columns: ", original_df.columns)
-
+for i, original_df in tqdm(enumerate(pd.read_json(file_path, lines=True, chunksize=10**4, dtype=str))):
     # Rename columns
     original_df.rename(columns={
         'major_category':   'field',
@@ -91,25 +68,13 @@ for i, original_df in tqdm(enumerate(pd.read_json(file_path, lines=True, chunksi
         'abs_categories':   'secondary_subfield'
     }, inplace=True)
     # ignore entries from ignored categories (i.e., primary subfield)
-    df = original_df[~original_df['field'].isin(IGNORED_CATEGOREIES)]
-    df = df[(df['version'] == 1)]
+    df = original_df[~original_df['field'].isin(GENERAL_CATEGORIES | IGNORED_CATEGOREIES)]
     if df['primary_subfield'].isin(CATEGORY_ALIASES.keys()).sum():
         ValueError("IMPLEMENT CODE TO CHANGE CATEGORY_ALIASES ")
 
-    # # preprocess the chunk here
-    # df_chunk = df.astype(str)
-    # # convert the DataFrame to a JSON string
-    # json_str = df_chunk.to_json(orient='records', lines=True)
-
-    # # write the JSON string to file
-    # f.write(json_str)
-
-    # # add a newline character after each chunk except the last one
-    # # if i != num_chunks - 1:
-    # f.write('\n')
     if i == 0:
-        df.astype(str).to_json(save_path, lines=True, orient="records")
+        df.to_json(save_path, lines=True, orient="records")
     else:
-        df.astype(str).to_json(save_path, lines=True, orient="records", mode='a')
+        df.to_json(save_path, lines=True, orient="records", mode='a')
 
 # %%

test_paper_id.py

@@ -0,0 +1,37 @@
+# script to test for any paper id overlap between the training and test set
+# to run:
+#   python test_paper_id.py
+
+import pandas as pd
+import re
+from tqdm import tqdm
+
+def test(dataset):
+    print(f"Checking dataset: {dataset}")
+    # load the training and test data
+    train_file = f"data/raw/{dataset}/train_{dataset}_cats_full.json"
+    # read json in chunks of 1000 rows, get the paper_id column and convert to list
+    train_ids = []
+    for df in tqdm(pd.read_json(train_file, lines=True, chunksize=1000, dtype=str)):
+        train_ids.extend(df['paper_id'].tolist())
+    test_file = f"data/raw/{dataset}/test_{dataset}_cats_full.json"
+    test_ids = []
+    for df in tqdm(pd.read_json(test_file, lines=True, chunksize=1000, dtype=str)):
+        test_ids.extend(df['paper_id'].tolist())
+
+    # check paper id formats
+    ARXIV_IDENTIFIER_PATTERN = r"^\d{2}(0[1-9]|1[0-2])\.\d{4,5}$"
+    OLD_ARXIV_IDENTIFIER_PATTERN = r"^[\w\-]+/\d{7}$"
+    assert not any([re.match(ARXIV_IDENTIFIER_PATTERN, pid) is None for pid in train_ids if '.' in pid])
+    assert not any([re.match(ARXIV_IDENTIFIER_PATTERN, pid) is None for pid in test_ids if '.' in pid])
+    assert not any([re.match(OLD_ARXIV_IDENTIFIER_PATTERN, pid) is None for pid in train_ids if '/' in pid])
+    assert not any([re.match(OLD_ARXIV_IDENTIFIER_PATTERN, pid) is None for pid in test_ids if '/' in pid])
+
+    # check for overlap
+    overlap = set(train_ids).intersection(set(test_ids))
+    assert len(overlap) == 0, f"Overlap found: {overlap}"
+
+if __name__ == "__main__":
+    test("major")
+    test("minor")
+    print("All tests passed!")