README

process.py

@@ -7,6 +7,9 @@ import requests
 
 from glob import glob
 import gdown
+k = 10  # the 3rd level negative-distance ranking
+m = 5  # the 3rd level negative-distance ranking
+top_n = 10  # threshold of positive pairs in the 1st and 2nd relation
 
 
 def wget(url, cache_dir: str = './cache', gdrive_filename: str = None):
@@ -50,7 +53,6 @@ def get_training_data(return_validation_set: bool = False):
     pairs: dictionary of list (positive pairs, negative pairs)
     {'1b': [[0.6, ('office', 'desk'), ..], [[-0.1, ('aaa', 'bbb'), ...]]
     """
-    top_n = 10
     cache_dir = 'cache'
     os.makedirs(cache_dir, exist_ok=True)
     remove_relation = None
@@ -98,7 +100,6 @@ def get_training_data(return_validation_set: bool = False):
         all_relation_type[relation_id] = relation_type
     parent = list(set([i[:-1] for i in all_relation_type.keys()]))
 
-
     # 1st level relation contrast (among parent relations)
     relation_pairs_1st = []
     for p in parent:
@@ -121,7 +122,8 @@ def get_training_data(return_validation_set: bool = False):
     relation_pairs_3rd = []
     for p in all_relation_type.keys():
         positive_pairs = positives[p]
-        negative_pairs = negatives[p]
+        negative_pairs = []
+        # negative_pairs = negatives[p]
         for n in all_relation_type.keys():
             if p == n:
                 continue
@@ -130,17 +132,18 @@ def get_training_data(return_validation_set: bool = False):
             "positives": positive_pairs, "negatives": negative_pairs, "relation_type": p, "level": "child"
         }]
 
+        negative_pairs = positive_pairs + negatives[p]
         for n, anchor in enumerate(positive_pairs):
+            if n > m:
+                continue
             for _n, posi in enumerate(positive_pairs):
-                if n < _n:
-                    negative_pairs = positive_pairs[_n+1:]
-                    if len(negative_pairs) > 0:
-                        relation_pairs_3rd += [{
-                            "positives": [(anchor, posi)],
-                            "negatives": [(anchor, neg) for neg in negative_pairs],
-                            "relation_type": p,
-                            "level": "child_prototypical"
-                        }]
+                if n < _n and len(negative_pairs) > _n + k:
+                    relation_pairs_3rd += [{
+                        "positives": [(anchor, posi)],
+                        "negatives": [(anchor, neg) for neg in negative_pairs[_n+k:]],
+                        "relation_type": p,
+                        "level": "child_prototypical"
+                    }]
 
     return relation_pairs_1st + relation_pairs_2nd + relation_pairs_3rd
 

semeval2012_relational_similarity_v3.py

@@ -4,7 +4,7 @@ import datasets
 logger = datasets.logging.get_logger(__name__)
 _DESCRIPTION = """[SemEVAL 2012 task 2: Relational Similarity](https://aclanthology.org/S12-1047/)"""
 _NAME = "semeval2012_relational_similarity_v3"
-_VERSION = "1.0.0"
+_VERSION = "1.1.0"
 _CITATION = """
 @inproceedings{jurgens-etal-2012-semeval,
     title = "{S}em{E}val-2012 Task 2: Measuring Degrees of Relational Similarity",