include training pairs (with sampled negative examples)

2018thresh10dev.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53c34495315acde41ca549eaa1ad02726e79c6f2aad46b77f84b11fb2459e666
+size 55066171

2018thresh10test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8089f88487de91ed778ecfed210f0f13488e3f670404c210ebd72ee3f1b268de
+size 36685709

2018thresh10train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2618a03dd0278017b987759f723cc7525d19ddc3c79247a187bcf0a2601663ed
+size 274277317

2018thresh20dev.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06c9b05cc92746020c8379d55d8738ca6923129ca8a15766df1687625d73cb30
+size 39942761

2018thresh20test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:feea8c3d46deb539e8004bc5193a15509cb3b352d93431dbe5131a5e4356b9d9
+size 26669268

2018thresh20train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebd8c98658d16de99412d9f52d3b98409c0bcefd36409378fe303d5dd61c9eba
+size 198305773

2018thresh5dev.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf2e6b8bf0b091f3cbcc3cfbdd831989e35be1f8331de0a2f0d217e574c189c4
+size 71983269

2018thresh5test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f654c09ae5f8ffbb80ddcb79c2be63e37b8ae184ef99c9555fb1c1625fedf1b7
+size 48195090

2018thresh5train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:866bad7a6fc1fcd477f87be9b856d289e47547878fcaeb000d1dba768a7b7468
+size 359260993

generate_wes_data.py

@@ -0,0 +1,76 @@
+from datasets import load_dataset
+import pandas as pd
+import numpy as np
+from tqdm import tqdm
+
+from collections import defaultdict
+from operator import itemgetter as ig
+from itertools import islice, chain, repeat
+from random import sample, choice, shuffle
+from gc import collect
+
+def generate_splits(subset, split=[0.75, 0.15, 0.1]):
+    assert abs(sum(split) - 1.0) < 0.0001
+    # get the data in dictionary form
+    groups = defaultdict(list)
+    ds = load_dataset('Exr0n/wiki-entity-similarity', subset, split='train')
+    ds = list(tqdm(ds, total=len(ds)))
+    for article, link in tqdm(map(ig('article', 'link_text'), ds), total=len(ds)):
+        groups[article].append(link)
+    del ds
+
+    # greedily allocate splits
+    order = sorted(groups.keys(), reverse=True, key=lambda e: groups[e])
+    splits = [[] for _ in split]
+    sizes = [0.001] * len(split)    # avoid div zero error
+    for group in order:
+        impoverished = np.argmax([ s - (x/sum(sizes)) for x, s in zip(sizes, split) ])
+        splits[impoverished].append(group)
+        sizes[impoverished] += len(groups[group])
+
+    sizes = [ int(x) for x in sizes ]
+    print('final sizes', sizes, [x/sum(sizes) for x in sizes])
+
+    # generate positive examples
+    ret = [ [[(k, t) for t in groups[k]] for k in keys] for keys in splits ]
+
+    # generate negative examples randomly (TODO: probably a more elegant swapping soln)
+    for i, keys in enumerate(splits):
+        for key in keys:
+            try:
+                got = sample(keys, len(groups[key])+1)
+                ret[i].append(
+                    [(key, choice(groups[k])) for k in got if k != key]
+                    [:len(groups[key])]
+                )
+            except ValueError:
+                raise ValueError("well frick one group is bigger than all the others combined. try sampling one at a time")
+
+    collect()
+    return [(chain(*s), chain(repeat(1, z), repeat(0, z))) for z, s in zip(sizes, ret)]
+
+
+if __name__ == '__main__':
+    for size in [5, 10, 20]:
+        x = generate_splits(subset='2018thresh' + str(size) + 'corpus')
+
+        for (data, labels), split in zip(x, ['train', 'dev', 'test']):
+            articles, lts = list(zip(*data))
+            df = pd.DataFrame({ 'article': articles, 'link_text': lts, 'is_same': list(labels) })
+            df = df.sample(frac=1).reset_index(drop=True)
+            df.to_csv('2018thresh' + str(size) + split + '.csv', index=False)
+            # print(df.head(30), df.tail(30))
+
+    # tests
+    # for data, labels in x[2:]:
+    #     data = list(data)
+    #     labels = list(labels)
+    #
+    #     assert sum(labels) * 2 == len(labels)
+    #     num = sum(labels)
+    #
+    #     before = [ a for a, _ in data[:num] ]
+    #     after  = [ a for a, _ in data[num:] ]
+    #     assert before == after
+    #
+    #     print(data[num:])

wiki-entity-similarity.py

@@ -12,33 +12,67 @@ _CITE = '''\
 }
 '''
 
+_HUGGINGFACE_REPO = "https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/"
+
 @dataclass
 class WikiEntitySimilarityConfig(datasets.BuilderConfig):
     """BuilderConfig for CSV."""
+    year: int = None
+    type: str = None
     threshhold: int = None
-    path: str = None
+    # path: str = None
 
 class WikiEntitySimilarity(datasets.GeneratorBasedBuilder):
     """WES: Learning semantic similarity from 6M names for 1M entities"""
     BUILDER_CONFIG_CLASS = WikiEntitySimilarityConfig
     BUILDER_CONFIGS = [
         WikiEntitySimilarityConfig(
-            name='2018thresh5',
-            description='min 5 inbound links, lowest quality',
+            name='2018thresh5corpus',
+            description='raw link corpus (all true): min 5 inbound links, lowest quality',
+            year=2018,
+            type='corpus',
+            threshhold=5,
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_5.csv"
+        ),
+        WikiEntitySimilarityConfig(
+            name='2018thresh10corpus',
+            description='raw link corpus (all true): min 10 inbound links, medium quality',
+            year=2018,
+            type='corpus',
+            threshhold=10,
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_10.csv"
+        ),
+        WikiEntitySimilarityConfig(
+            name='2018thresh20corpus',
+            description='raw link corpus (all true): min 20 inbound links, high quality',
+            year=2018,
+            type='corpus',
+            threshhold=20,
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_20.csv"
+        ),
+        WikiEntitySimilarityConfig(
+            name='2018thresh5pairs',
+            description='training pairs based on min 5 inbound links, lowest quality',
+            year=2018,
+            type='pairs',
             threshhold=5,
-            path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_5.csv"
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/2018thresh5"
         ),
         WikiEntitySimilarityConfig(
-            name='2018thresh10',
-            description='min 10 inbound links, medium quality',
+            name='2018thresh10pairs',
+            description='training pairs based on min 10 inbound links, medium quality',
+            year=2018,
+            type='pairs',
             threshhold=10,
-            path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_10.csv"
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/2018thresh10"
         ),
         WikiEntitySimilarityConfig(
-            name='2018thresh20',
-            description='min 20 inbound links, high quality',
+            name='2018thresh20pairs',
+            description='training pairs based on min 20 inbound links, high quality',
+            year=2018,
+            type='pairs',
             threshhold=20,
-            path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/link_synonyms-2018-thresh_20.csv"
+            # path="https://huggingface.co/datasets/Exr0n/wiki-entity-similarity/resolve/main/2018thresh20"
         ),
     ]
 
@@ -56,12 +90,25 @@ class WikiEntitySimilarity(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        filepath = dl_manager.download(self.config.path)
-        return [ datasets.SplitGenerator(name=datasets.Split.TRAIN,
-                                         gen_kwargs={ 'filepath': filepath }) ]
+        path = _HUGGINGFACE_REPO + f"{self.config.year}thresh{self.config.threshhold}"
+        if self.config.type == 'corpus':
+            filepath = dl_manager.download(path + 'corpus.csv')
+            return [ datasets.SplitGenerator(name=datasets.Split.TRAIN,
+                                             gen_kwargs={ 'path': filepath }) ]
+        elif self.config.type == 'pairs':
+            ret = []
+            for n, e in zip(['train', 'dev', 'test'],
+                            [datasets.Split.TRAIN,
+                             datasets.Split.VALIDATION,
+                             datasets.Split.TEST]):
+                fp = dl_manager.download(path + n + '.csv')
+                ret.append( datasets.SplitGenerator(name=e, gen_kwargs={ 'path': fp }) )
+            return ret
+        else:
+            raise ValueError(f"invalid dataset type '{self.config.type}', expected 'corpus' for raw links or 'pairs' for trainable pairs with negative examples")
 
-    def _generate_examples(self, filepath):
-        with open(filepath, 'r') as rf:
+    def _generate_examples(self, path):
+        with open(path, 'r') as rf:
             reader = csv.DictReader(rf)
             for i, row in enumerate(reader):
                 yield i, row