runnable v1

.gitignore

@@ -7,3 +7,4 @@ dataset/FB15k-237/masked_*.txt
 dataset/FB15k-237/cached_*.pkl
 **/__pycache__/
 **/.DS_Store
+nohup.out

.vscode/launch.json

@@ -0,0 +1,38 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Python: Current File",
+            "type": "python",
+            "request": "launch",
+            "program": "${file}",
+            "console": "integratedTerminal",
+            "justMyCode": true,
+            "args": [
+                "--gpus", "1", 
+                "--max_epochs=16",  
+                "--num_workers=32", 
+                "--model_name_or_path",  "bert-base-uncased",
+                "--accumulate_grad_batches", "1", 
+                "--model_class", "BertKGC",
+                "--batch_size", "64",
+                "--checkpoint", "/kg_374/Relphormer/pretrain/output/FB15k-237/epoch=15-step=19299-Eval/hits10=0.96.ckpt",
+                "--pretrain", "0",
+                "--bce", "0",
+                "--check_val_every_n_epoch", "1",
+                "--overwrite_cache",
+                "--data_dir", "dataset/FB15k-237", 
+                "--eval_batch_size", "128",
+                "--max_seq_length", "128",
+                "--lr", "3e-5",
+                "--max_triplet", "64",
+                "--add_attn_bias", "True",
+                "--use_global_node", "True",
+                "--fast_dev_run", "True",
+            ]
+        }
+    ]
+}

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2021 ZJUNLP
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

QA/__init__.py

@@ -0,0 +1,6 @@
+
+
+from transformers import BartForConditionalGeneration, T5ForConditionalGeneration, GPT2LMHeadModel
+
+from .model import *
+

QA/fbqa/relphormer-filtered/dev-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3994,
+	"#kept_examples": 3994,
+	"#mappable_examples": 743,
+	"#multiple_answer_examples": 2
+}

QA/fbqa/relphormer-filtered/test-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3996,
+	"#kept_examples": 3996,
+	"#mappable_examples": 755,
+	"#multiple_answer_examples": 0
+}

QA/fbqa/relphormer-filtered/train-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 20358,
+	"#kept_examples": 20358,
+	"#mappable_examples": 3713,
+	"#multiple_answer_examples": 4
+}

QA/fbqa/relphormer/dev-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3994,
+	"#kept_examples": 3994,
+	"#mappable_examples": 743,
+	"#multiple_answer_examples": 2
+}

QA/fbqa/relphormer/test-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3996,
+	"#kept_examples": 3996,
+	"#mappable_examples": 755,
+	"#multiple_answer_examples": 0
+}

QA/fbqa/relphormer/train-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 20358,
+	"#kept_examples": 20358,
+	"#mappable_examples": 3713,
+	"#multiple_answer_examples": 4
+}

QA/hitter-bert.py

@@ -0,0 +1,403 @@
+import json
+import math
+import argparse
+from pathlib import Path
+
+from transformers import BertTokenizer, BertForMaskedLM, AdamW, get_linear_schedule_with_warmup, AutoConfig
+
+import torch
+from torch import device, nn
+from torch.utils.data import DataLoader, Dataset
+
+import pytorch_lightning as pl
+from pytorch_lightning.loggers import TensorBoardLogger
+from pytorch_lightning.callbacks import LearningRateMonitor, ModelCheckpoint
+from pytorch_lightning.utilities.seed import seed_everything
+from transformers.tokenization_bert import BertTokenizerFast
+
+from kge.model import KgeModel
+from kge.util.io import load_checkpoint
+from kge.util import sc
+# from relphormer.lit_models import TransformerLitModel
+from relphormer.models import BertKGC
+# from relphormer.data import KGC
+import os
+
+os.environ['CUDA_VISIBLE_DEVICES']='4'
+
+MODEL = 'bert-base-uncased'
+tokenizer = BertTokenizer.from_pretrained(MODEL)
+
+
+class FBQADataset(Dataset):
+
+    def __init__(self, file_dir):
+        self.examples = json.load(Path(file_dir).open("rb"))
+
+    def __len__(self):
+        return len(self.examples)
+
+    def __getitem__(self, idx):
+        if torch.is_tensor(idx):
+            idx = idx.tolist()
+        return self.examples[idx]
+
+
+def fbqa_collate(samples):
+    questions = []
+    answers = []
+    answer_ids = []
+    entities = []
+    entity_names = []
+    relations = []
+    for item in samples:
+        q = item["RawQuestion"] + "[MASK]" * len(item["AnswerEntity"]) + "."
+        questions.append(q)
+        answers.append(item["AnswerEntity"])
+        answer_ids.append(item["AnswerEntityID"])
+        entities.append(item["TopicEntityID"])
+        entity_names.append(item["TopicEntityName"])
+        relations.append(item["RelationID"])
+        
+    questions = tokenizer(questions, return_tensors='pt', padding=True)
+    entity_names = tokenizer(entity_names, add_special_tokens=False)
+    answers, answers_lengths = sc.pad_seq_of_seq(answers)
+    answers = torch.LongTensor(answers)
+    answers_lengths = torch.LongTensor(answers_lengths)
+    answer_ids = torch.LongTensor(answer_ids)
+
+    input_ids = questions['input_ids']
+    masked_labels = torch.ones_like(input_ids) * -100
+    masked_labels[input_ids == tokenizer.mask_token_id] = answers[answers != 0]
+    entity_mask = torch.zeros_like(input_ids).bool()
+    entity_span_index = input_ids.new_zeros((len(input_ids), 2))
+    for i, e_tokens in enumerate(entity_names['input_ids']):
+        q_tokens = input_ids[i].tolist()
+        for s_index in range(len(q_tokens) - len(e_tokens)):
+            if all([e_token == q_tokens[s_index + j] for j, e_token in enumerate(e_tokens)]):
+                entity_mask[i][s_index:s_index + len(e_tokens)] = True
+                entity_span_index[i][0] = s_index
+                entity_span_index[i][1] = s_index + len(e_tokens) - 1
+                break
+
+    entities = torch.LongTensor(entities)
+    relations = torch.LongTensor(relations)
+
+    return questions.data, masked_labels, answers, answers_lengths, answer_ids, entities, relations, entity_mask, entity_span_index
+
+
+class SelfOutput(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+        self.dropout = nn.Dropout(config.hidden_dropout_prob)
+
+    def forward(self, hidden_states, input_tensor):
+        hidden_states = self.dense(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+        hidden_states = self.LayerNorm(hidden_states + input_tensor)
+        return hidden_states
+
+
+class CrossAttention(nn.Module):
+    def __init__(self, config, ctx_hidden_size):
+        super().__init__()
+        self.self = CrossAttentionInternal(config, ctx_hidden_size)
+        self.output = SelfOutput(config)
+        self.config = config
+        self.apply(self._init_weights)
+
+    def _init_weights(self, module):
+        """ Initialize the weights """
+        if isinstance(module, (nn.Linear, nn.Embedding)):
+            # Slightly different from the TF version which uses truncated_normal for initialization
+            # cf https://github.com/pytorch/pytorch/pull/5617
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+        elif isinstance(module, nn.LayerNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+        if isinstance(module, nn.Linear) and module.bias is not None:
+            module.bias.data.zero_()
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        output_attentions=False,
+    ):
+        self_outputs = self.self(
+            hidden_states,
+            attention_mask,
+            head_mask,
+            encoder_hidden_states,
+            encoder_attention_mask,
+            output_attentions,
+        )
+        attention_output = self.output(self_outputs[0], hidden_states)
+        outputs = (attention_output,) + self_outputs[1:]  # add attentions if we output them
+        return outputs
+
+
+class CrossAttentionInternal(nn.Module):
+    def __init__(self, config, ctx_hidden_size):
+        super().__init__()
+        if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"):
+            raise ValueError(
+                "The hidden size (%d) is not a multiple of the number of attention "
+                "heads (%d)" % (config.hidden_size, config.num_attention_heads)
+            )
+
+        self.num_attention_heads = config.num_attention_heads
+        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)
+        self.all_head_size = self.num_attention_heads * self.attention_head_size
+
+        self.query = nn.Linear(config.hidden_size, self.all_head_size)
+        self.key = nn.Linear(ctx_hidden_size, self.all_head_size)
+        self.value = nn.Linear(ctx_hidden_size, self.all_head_size)
+
+        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
+
+    def transpose_for_scores(self, x):
+        new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size)
+        x = x.view(*new_x_shape)
+        return x.permute(0, 2, 1, 3)
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        output_attentions=False,
+    ):
+        mixed_query_layer = self.query(hidden_states)
+
+        # If this is instantiated as a cross-attention module, the keys
+        # and values come from an encoder; the attention mask needs to be
+        # such that the encoder's padding tokens are not attended to.
+        mixed_key_layer = self.key(encoder_hidden_states)
+        mixed_value_layer = self.value(encoder_hidden_states)
+        attention_mask = encoder_attention_mask
+
+        query_layer = self.transpose_for_scores(mixed_query_layer)
+        key_layer = self.transpose_for_scores(mixed_key_layer)
+        value_layer = self.transpose_for_scores(mixed_value_layer)
+
+        # Take the dot product between "query" and "key" to get the raw attention scores.
+        attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))
+        attention_scores = attention_scores / math.sqrt(self.attention_head_size)
+        if attention_mask is not None:
+            # Apply the attention mask is (precomputed for all layers in BertModel forward() function)
+            attention_scores = attention_scores + attention_mask
+
+        # Normalize the attention scores to probabilities.
+        attention_probs = nn.Softmax(dim=-1)(attention_scores)
+
+        # This is actually dropping out entire tokens to attend to, which might
+        # seem a bit unusual, but is taken from the original Transformer paper.
+        attention_probs = self.dropout(attention_probs)
+
+        # Mask heads if we want to
+        if head_mask is not None:
+            attention_probs = attention_probs * head_mask
+
+        context_layer = torch.matmul(attention_probs, value_layer)
+
+        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
+        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)
+        context_layer = context_layer.view(*new_context_layer_shape)
+
+        outputs = (context_layer, nn.Softmax(dim=-1)(attention_scores)) if output_attentions else (context_layer,)
+        return outputs
+
+
+class CrossTrmFinetuner(pl.LightningModule):
+    def __init__(self, hparams, bertmodel):
+        super().__init__()
+        self._hparams = hparams
+
+        self.lr = hparams['lr']
+        self.weight_decay = hparams['weight_decay']
+
+        self.kg_dim = 320
+        # self.bert = BertForMaskedLM.from_pretrained(MODEL)
+        self.bert = bertmodel
+
+        if self._hparams['use_hitter']:
+            self.kg_layer_num = 10
+            self.cross_attentions = nn.ModuleList([CrossAttention(self.bert.config, self.kg_dim)
+                                                   for _ in range(self.kg_layer_num)])
+            checkpoint = load_checkpoint('local/best/20200812-174221-trmeh-fb15k237-best/checkpoint_best.pt')
+            self.hitter = KgeModel.create_from(checkpoint)
+
+    def forward(self, batch):
+        sent_input, masked_labels, batch_labels, label_lens, answer_ids, s, p, entity_mask, entity_span_index = batch
+
+        if self._hparams['use_hitter']:
+            # kg_masks: [bs, 1, 1, length]
+            # kg_embeds: nlayer*[bs, length, dim]
+            kg_embeds, kg_masks = self.hitter('get_hitter_repr', s, p)
+            kg_attentions = [None] * 2 + [(self.cross_attentions[i], kg_embeds[(i + 2) // 2], kg_masks)
+                                          for i in range(self.kg_layer_num)]
+        else:
+            kg_attentions = []
+
+        out = self.bert(kg_attentions=kg_attentions,
+                        output_attentions=True,
+                        output_hidden_states=True,
+                        return_dict=True,
+                        labels=masked_labels,
+                        **sent_input,
+                        )
+
+        return out
+
+    def training_step(self, batch, batch_idx):
+        output = self(batch)
+        loss = output.loss
+        self.log('train_loss', loss, on_epoch=True, prog_bar=True)
+        return {'loss': loss}
+
+    def validation_step(self, batch, batch_idx):
+        batch_inputs, masked_labels, batch_labels, label_lens, answer_ids, s, p, entity_mask, _ = batch
+        output = self(batch)
+        input_tokens = batch_inputs["input_ids"].clone()
+
+        logits = output.logits[masked_labels != -100]
+        probs = logits.softmax(dim=-1)
+        values, predictions = probs.topk(1)
+        hits = []
+        now_pos = 0
+        for sample_i, label_length in enumerate(label_lens.tolist()):
+            failed = False
+            for i in range(label_length):
+                if (predictions[now_pos + i] == batch_labels[sample_i][i]).sum() != 1:
+                    failed = True
+                    break
+            hits += [1] if not failed else [0]
+            now_pos += label_length
+        hits = torch.tensor(hits)
+        input_tokens[input_tokens == tokenizer.mask_token_id] = predictions.flatten()
+        pred_strings = [str(hits[i].item()) + ' ' + tokenizer.decode(input_tokens[i], skip_special_tokens=True)
+                        for i in range(input_tokens.size(0))]
+
+        return {'val_loss': output.loss,
+                'val_acc': hits.float(),
+                'pred_strings': pred_strings}
+
+    def validation_epoch_end(self, outputs):
+        avg_loss = torch.stack([x['val_loss'] for x in outputs]).mean()
+        avg_val_acc = torch.cat([x['val_acc'] for x in outputs]).mean().to(avg_loss.device)
+
+        if self.global_rank == 0:
+            tensorboard = self.logger.experiment
+            tensorboard.add_text('pred', '\n\n'.join(sum([x['pred_strings'] for x in outputs], [])), self.global_step)
+
+        self.log('avg_loss', avg_loss, on_epoch=True, prog_bar=True, sync_dist=True)
+        self.log('avg_val_acc', avg_val_acc, on_epoch=True, prog_bar=True, sync_dist=True)
+        return {'val_loss': avg_loss}
+
+    def train_dataloader(self):
+        return DataLoader(FBQADataset(self._hparams['train_dataset']),
+                          self._hparams['batch_size'],
+                          shuffle=True,
+                          collate_fn=fbqa_collate,
+                          num_workers=0)
+
+    def val_dataloader(self):
+        return DataLoader(FBQADataset(self._hparams['val_dataset']),
+                          1,
+                          shuffle=False,
+                          collate_fn=fbqa_collate,
+                          num_workers=0)
+
+    def test_dataloader(self):
+        return DataLoader(FBQADataset(self._hparams['test_dataset']),
+                          1,
+                          shuffle=False,
+                          collate_fn=fbqa_collate,
+                          num_workers=0)
+
+    def configure_optimizers(self):
+        no_decay = ['bias', 'LayerNorm.weight']
+        no_fine_tune = ['cross_attentions']
+        pgs = [{'params': [p for n, p in self.named_parameters() if not any(nd in n for nd in no_decay) and not any([i in n for i in no_fine_tune])],
+                'weight_decay': 0.01},
+               {'params': [p for n, p in self.named_parameters() if any(nd in n for nd in no_decay) and not any([i in n for i in no_fine_tune])],
+                'weight_decay': 0.0}]
+        if self._hparams['use_hitter']:
+            pgs.append({'params': self.cross_attentions.parameters(), 'lr': 5e-5, 'weight_decay': 0.01})
+        # bert_optimizer = AdamW(pgs, lr=3e-5, weight_decay=1e-2)
+        bert_optimizer = AdamW(pgs, lr=self.lr, weight_decay=self.weight_decay)
+        bert_scheduler = {
+            'scheduler': get_linear_schedule_with_warmup(bert_optimizer, self._hparams['max_steps'] // 10, self._hparams['max_steps']),
+            'interval': 'step',
+            'monitor': None
+        }
+        return [bert_optimizer], [bert_scheduler]
+    
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--exp_name", default='default', nargs='?', help="Name of the experiment")
+    parser.add_argument('--dataset', choices=['fbqa', 'webqsp'], default='fbqa', help="fbqa or webqsp")
+    parser.add_argument('--filtered', default=False, action='store_true', help="Filtered or not")
+    parser.add_argument('--hitter', default=False, action='store_true', help="Use pretrained HittER or not")
+    parser.add_argument('--relphormer', default=False, action='store_true', help="Use pretrained relphormer or not")
+    parser.add_argument('--seed', default=333, type=int, help='Seed number')
+    parser.add_argument('--lr', default=3e-5, type=float, help='learning rate')
+    parser.add_argument('--weight_decay', default=1e-2, type=float, help='weight decay')
+    args = parser.parse_args()
+    seed_everything(args.seed)
+
+    QA_DATASET = args.dataset
+    if args.filtered and args.relphormer:
+        SUBSET = 'relphormer-filtered'
+    elif not args.filtered and args.relphormer:
+        SUBSET = 'relphormer'
+    elif args.filtered and not args.relphormer:
+        SUBSET = 'fb15k237-filtered'
+    else:
+        SUBSET = 'fb15k237'
+
+    hparams = {
+        'use_hitter': args.hitter,
+        'relphormer': args.relphormer,
+        'lr': args.lr,
+        'weight_decay': args.weight_decay,
+        'batch_size': 16,
+        'max_epochs': 20,
+        'train_dataset': f'data/{QA_DATASET}/{SUBSET}/train.json',
+        'val_dataset': f'data/{QA_DATASET}/{SUBSET}/test.json',
+        'test_dataset': f'data/{QA_DATASET}/{SUBSET}/test.json',
+    }
+
+    if hparams['relphormer']:
+        MODEL = "./local/relphormer/"
+        config = AutoConfig.from_pretrained(MODEL)
+        bertmodel = BertForMaskedLM.from_pretrained(MODEL, config=config)
+        model = CrossTrmFinetuner(hparams, bertmodel=bertmodel)
+    else:
+        bertmodel = BertForMaskedLM.from_pretrained(MODEL)
+        model = CrossTrmFinetuner(hparams, bertmodel=bertmodel)
+    model.hparams['max_steps'] = (len(model.train_dataloader().dataset) // hparams['batch_size'] + 1) * hparams['max_epochs']
+    base_path = '/tmp/hitbert-paper'
+    logger = TensorBoardLogger(base_path, args.exp_name)
+    checkpoint_callback = ModelCheckpoint(
+            monitor='avg_val_acc',
+            dirpath=base_path + '/' + args.exp_name,
+            filename='{epoch:02d}-{avg_val_acc:.3f}',
+            save_top_k=1,
+            mode='max')
+    trainer = pl.Trainer(gpus=1, accelerator="ddp",
+                         max_epochs=hparams['max_epochs'], max_steps=model.hparams['max_steps'],
+                         checkpoint_callback=True,
+                         gradient_clip_val=1.0, logger=logger,
+                         callbacks=[LearningRateMonitor(), checkpoint_callback])
+    trainer.fit(model)
+    print("QA Task End!")

QA/model.py

@@ -0,0 +1,8 @@
+# from transformers.models.bert.modeling_bert import BertForMaskedLM
+from models.huggingface_relformer import BertForMaskedLM
+class BertKGC(BertForMaskedLM):
+
+    @staticmethod
+    def add_to_argparse(parser):
+        parser.add_argument("--pretrain", type=int, default=0, help="")
+        return parser

QA/scripts/relphormer_fbqa.sh

@@ -0,0 +1,10 @@
+for SEED in  111 222 333 444 555 666 777 888 999
+do
+# echo ${LR} ${WD}
+python hitter-bert.py --dataset fbqa \
+        --relphormer \
+        --seed ${SEED} \
+        --exp_name relphormer-fbqa \
+        --lr 3e-5 \
+        --weight_decay 1e-2
+done

QA/scripts/relphormer_fbqa_filtered.sh

@@ -0,0 +1,13 @@
+
+for SEED in  111 222 333 444 555 666 777 888 999
+do
+
+# echo ${LR} ${WD}
+python hitter-bert.py --dataset fbqa \
+        --relphormer \
+        --filtered \
+        --seed ${SEED} \
+        --exp_name relphormer-filtered-fbqa \
+        --lr 3e-5 \
+        --weight_decay 1e-2
+done

QA/scripts/relphormer_webqsp.sh

@@ -0,0 +1,10 @@
+
+for SEED in  222 333 444 555 666 777 888 999
+do
+python hitter-bert.py --dataset webqsp \
+        --relphormer \
+        --seed ${SEED} \
+        --exp_name relphormer-webqsp \
+        --lr 3e-5 \
+        --weight_decay 1e-2
+done

QA/scripts/relphormer_webqsp_filtered.sh

@@ -0,0 +1,12 @@
+
+for SEED in  111 222 333 444 555 666 777 888 999
+do
+# echo ${LR} ${WD}
+python hitter-bert.py --dataset webqsp \
+        --relphormer \
+        --filtered \
+        --seed ${SEED} \
+        --exp_name relphormer-filtered-webqsp \
+        --lr 3e-5 \
+        --weight_decay 1e-2
+done

QA/webqsp/relphormer-filtered/test-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 1639,
+	"#kept_examples": 484,
+	"#mappable_examples": 484,
+	"#multiple_answer_examples": 800
+}

QA/webqsp/relphormer-filtered/train-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3098,
+	"#kept_examples": 850,
+	"#mappable_examples": 850,
+	"#multiple_answer_examples": 1437
+}

QA/webqsp/relphormer/test-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 1639,
+	"#kept_examples": 1582,
+	"#mappable_examples": 484,
+	"#multiple_answer_examples": 800
+}

QA/webqsp/relphormer/train-stat.json

@@ -0,0 +1,6 @@
+{
+	"#examples": 3098,
+	"#kept_examples": 2997,
+	"#mappable_examples": 850,
+	"#multiple_answer_examples": 1437
+}

README.md

@@ -0,0 +1,115 @@
+# Relphormer
+
+Code for the paper: "Relphormer: Relational Graph Transformer for Knowledge Graph Representations".
+
+> Transformers have achieved remarkable performance in widespread fields, including natural language processing, computer vision and graph mining. However, vanilla Transformer architectures have not yielded promising improvements in the Knowledge Graph (KG) representations, where the translational distance paradigm dominates this area. Note that vanilla Transformer architectures struggle to capture the intrinsically heterogeneous semantic and structural information of knowledge graphs. To this end, we propose a new variant of Transformer for knowledge graph representations dubbed Relphormer. Specifically, we introduce Triple2Seq which can dynamically sample contextualized sub-graph sequences as the input to alleviate the heterogeneity issue. We propose a novel structure-enhanced self-attention mechanism to encode the relational information and keep the globally semantic information among sub-graphs. Moreover, we propose masked knowledge modeling as a new paradigm for knowledge graph representation learning. We apply Relphormer to three tasks, namely, knowledge graph completion, KG-based question answering and KG-based recommendation for evaluation. Experimental results show that Relphormer can obtain better performance on benchmark datasets compared with baselines.
+
+
+# Model Architecture
+
+<div align=center>
+<img src="./resource/model.png" width="85%" height="75%" />
+</div>
+ 
+
+The model architecture of Relphormer. 
+The contextualized sub-graph is sampled with Triple2Seq, and then it will be converted into sequences while maintaining its sub-graph structure.
+Next, we conduct masked knowledge modeling, which randomly masks the nodes in the center triple in the contextualized sub-graph sequences.
+For the transformer architecture, we design a novel structure-enhanced mechanism to preserve the structure feature.
+Finally, we utilize our pre-trained KG transformer for KG-based downstream tasks. 
+
+# Environments
+
+- python (3.8.13)
+- cuda(11.2)
+- Ubuntu-18.04.6 (4.15.0-156-generic)
+
+# Requirements
+
+To run the codes, you need to install the requirements:
+```
+pip install -r requirements.txt
+```
+
+The expected structure of files is:
+
+```
+ ── Relphormer
+    ├── data
+    ├── dataset
+    │   ├── FB15k-237
+    │   ├── WN18RR
+    │   ├── umls
+    │   ├── create_neighbor.py
+    ├── lit_models
+    │   ├── _init_.py
+    │   ├── base.py
+    │   ├── transformer.py
+    │   └── utils.py
+    ├── models
+    │   ├── _init_.py
+    │   ├── huggingface_relformer.py
+    │   ├── model.py
+    │   └── utils.py    
+    ├── resource
+    │   └── model.png    
+    ├── scripts
+    │   ├── fb15k-237
+    │   ├── wn18rr
+    │   └── umls
+    ├── QA
+    ├── logs
+    ├── main.py
+    └── requirements.txt
+```
+
+# How to run
+
+## KGC Task
+
+### Generate Masked Neighbors
+
+- Use the command below to generate the masked neighbors.
+```shell
+>> cd dataset
+>> python create_neighbor.py --dataset xxx  # like python create_neighbor.py --dataset umls
+```
+
+### Entity Embedding Initialization
+
+- Then use the command below to add entities to BERT and initialize the entity embedding layer to be used in the later training. For other datasets `FB15k-237`  and `WN18RR` ,  just replace the dataset name with  `fb15k-237` and  `wn18rr` will be fine.
+
+```shell
+>> cd pretrain
+>> mkdir logs
+>> bash scripts/pretrain_umls.sh
+>> tail -f -n 2000 logs/pretrain_umls.log
+```
+
+The pretrained models are saved in the `Relphormer/pretrain/output` directory.
+
+### Entity Prediction
+
+- Next use the command below to train the model to predict the correct entity in the masked position. Same as above for other datasets.
+
+```shell
+>> cd Relphormer
+>> mkdir logs
+>> bash scripts/umls/umls.sh
+>> tail -f -n 2000 logs/train_umls.log
+```
+
+The trained models are saved in the `Relphormer/output` directory.
+
+## QA Task
+The experimental settings in QA follow the [Hitter](https://arxiv.org/pdf/2008.12813.pdf) experimental settings, and the environment installation can be done by referring to [GitHub](https://github.com/microsoft/HittER). We only modified **hitter-best.py** to fit our model.
+
+- The relphormer model used by QA can be downloaded [here](https://drive.google.com/file/d/1FK_A_kFq1ECoNm75RfkcvYv8rZiJL1Bw/view?usp=sharing).
+
+```shell
+>> cd QA
+>> sh scripts/relphormer_fbqa.sh
+>> sh scripts/relphormer_fbqa_filtered.sh 
+>> sh scripts/relphormer_webqsp.sh
+>> sh scripts/relphormer_webqsp_filtered.sh
+```

config/log_config.json

@@ -1,24 +0,0 @@
-{
-    "version": 1,
-    "disable_existing_loggers": false,
-    "formatters": {
-        "simple": {
-            "format": "%(asctime)s - %(name)s - [%(levelname)s] - %(message)s"
-        }
-    },
-    "handlers": {
-        "file_handler": {
-            "class": "logging.FileHandler",
-            "level": "DEBUG",
-            "formatter": "simple",
-            "filename": "python_logging.log",
-            "encoding": "utf8"
-        }
-    },
-    "root": {
-        "level": "DEBUG",
-        "handlers": [
-            "file_handler"
-        ]
-    }
-}

data/__init__.py

@@ -0,0 +1,2 @@
+from .data_module import KGC
+from .processor import convert_examples_to_features, KGProcessor

data/algos.pyx

@@ -0,0 +1,63 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+
+import cython
+from cython.parallel cimport prange, parallel
+cimport numpy
+import numpy
+
+def floyd_warshall(adjacency_matrix):
+
+    (nrows, ncols) = adjacency_matrix.shape
+    assert nrows == ncols
+    cdef unsigned int n = nrows
+
+    adj_mat_copy = adjacency_matrix.astype(long, order='C', casting='safe', copy=True)
+    assert adj_mat_copy.flags['C_CONTIGUOUS']
+    cdef numpy.ndarray[long, ndim=2, mode='c'] M = adj_mat_copy
+    cdef numpy.ndarray[long, ndim=2, mode='c'] path = numpy.zeros([n, n], dtype=numpy.int64)
+
+    cdef unsigned int i, j, k
+    cdef long M_ij, M_ik, cost_ikkj
+    cdef long* M_ptr = &M[0,0]
+    cdef long* M_i_ptr
+    cdef long* M_k_ptr
+
+    # set unreachable nodes distance to 510
+    for i in range(n):
+        for j in range(n):
+            if i == j:
+                M[i][j] = 0
+            elif M[i][j] == 0:
+                M[i][j] = 510
+
+    # floyed algo
+    for k in range(n):
+        M_k_ptr = M_ptr + n*k
+        for i in range(n):
+            M_i_ptr = M_ptr + n*i
+            M_ik = M_i_ptr[k]
+            for j in range(n):
+                cost_ikkj = M_ik + M_k_ptr[j]
+                M_ij = M_i_ptr[j]
+                if M_ij > cost_ikkj:
+                    M_i_ptr[j] = cost_ikkj
+                    path[i][j] = k
+
+    # set unreachable path to 510
+    for i in range(n):
+        for j in range(n):
+            if M[i][j] >= 510:
+                path[i][j] = 510
+                M[i][j] = 510
+
+    return M, path
+
+
+def get_all_edges(path, i, j):
+    cdef unsigned int k = path[i][j]
+    if k == 0:
+        return []
+    else:
+        return get_all_edges(path, i, k) + [k] + get_all_edges(path, k, j)
+

data/base_data_module.py

@@ -0,0 +1,71 @@
+"""Base DataModule class."""
+from pathlib import Path
+from typing import Dict
+import argparse
+import os
+
+import pytorch_lightning as pl
+from torch.utils.data import DataLoader
+
+
+class Config(dict):
+    def __getattr__(self, name):
+        return self.get(name)
+
+    def __setattr__(self, name, val):
+        self[name] = val
+
+
+BATCH_SIZE = 8
+NUM_WORKERS = 8
+
+
+class BaseDataModule(pl.LightningDataModule):
+    """
+    Base DataModule.
+    Learn more at https://pytorch-lightning.readthedocs.io/en/stable/datamodules.html
+    """
+
+    def __init__(self, args: argparse.Namespace = None) -> None:
+        super().__init__()
+        self.args = Config(vars(args)) if args is not None else {}
+        self.batch_size = self.args.get("batch_size", BATCH_SIZE)
+        self.num_workers = self.args.get("num_workers", NUM_WORKERS)
+
+
+    @staticmethod
+    def add_to_argparse(parser):
+        parser.add_argument(
+            "--batch_size", type=int, default=BATCH_SIZE, help="Number of examples to operate on per forward step."
+        )
+        parser.add_argument(
+            "--num_workers", type=int, default=0, help="Number of additional processes to load data."
+        )
+        parser.add_argument(
+            "--dataset", type=str, default="./dataset/NELL", help="Number of additional processes to load data."
+        )
+        return parser
+
+    def prepare_data(self):
+        """
+        Use this method to do things that might write to disk or that need to be done only from a single GPU in distributed settings (so don't set state `self.x = y`).
+        """
+        pass
+
+    def setup(self, stage=None):
+        """
+        Split into train, val, test, and set dims.
+        Should assign `torch Dataset` objects to self.data_train, self.data_val, and optionally self.data_test.
+        """
+        self.data_train = None
+        self.data_val = None
+        self.data_test = None
+
+    def train_dataloader(self):
+        return DataLoader(self.data_train, shuffle=True, batch_size=self.batch_size, num_workers=self.num_workers, pin_memory=True)
+
+    def val_dataloader(self):
+        return DataLoader(self.data_val, shuffle=False, batch_size=self.batch_size, num_workers=self.num_workers, pin_memory=True)
+
+    def test_dataloader(self):
+        return DataLoader(self.data_test, shuffle=False, batch_size=self.batch_size, num_workers=self.num_workers, pin_memory=True)

data/data_module.py

@@ -0,0 +1,198 @@
+from dataclasses import dataclass
+from typing import Any, Callable, Dict, List, NewType, Optional, Tuple, Union
+from enum import Enum
+import torch
+
+from torch.utils.data import DataLoader
+from transformers import AutoTokenizer, BertTokenizer
+# from transformers.configuration_bert import BertTokenizer, BertTokenizerFast
+from transformers.tokenization_utils_base import (BatchEncoding,
+                                                  PreTrainedTokenizerBase)
+
+from .base_data_module import BaseDataModule
+from .processor import KGProcessor, get_dataset, getNegativeEntityId
+import transformers
+transformers.logging.set_verbosity_error()
+
+class ExplicitEnum(Enum):
+    """
+    Enum with more explicit error message for missing values.
+    """
+
+    @classmethod
+    def _missing_(cls, value):
+        raise ValueError(
+            f"{value} is not a valid {cls.__name__}, please select one of {list(cls._value2member_map_.keys())}"
+        )
+
+class PaddingStrategy(ExplicitEnum):
+    """
+    Possible values for the ``padding`` argument in :meth:`PreTrainedTokenizerBase.__call__`. Useful for tab-completion
+    in an IDE.
+    """
+
+    LONGEST = "longest"
+    MAX_LENGTH = "max_length"
+    DO_NOT_PAD = "do_not_pad"
+
+import numpy as np
+
+@dataclass
+class DataCollatorForSeq2Seq:
+    """
+    Data collator that will dynamically pad the inputs received, as well as the labels.
+
+    Args:
+        tokenizer (:class:`~transformers.PreTrainedTokenizer` or :class:`~transformers.PreTrainedTokenizerFast`):
+            The tokenizer used for encoding the data.
+        model (:class:`~transformers.PreTrainedModel`):
+            The model that is being trained. If set and has the `prepare_decoder_input_ids_from_labels`, use it to
+            prepare the `decoder_input_ids`
+
+            This is useful when using `label_smoothing` to avoid calculating loss twice.
+        padding (:obj:`bool`, :obj:`str` or :class:`~transformers.file_utils.PaddingStrategy`, `optional`, defaults to :obj:`True`):
+            Select a strategy to pad the returned sequences (according to the model's padding side and padding index)
+            among:
+
+            * :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
+              sequence is provided).
+            * :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
+              maximum acceptable input length for the model if that argument is not provided.
+            * :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
+              different lengths).
+        max_length (:obj:`int`, `optional`):
+            Maximum length of the returned list and optionally padding length (see above).
+        pad_to_multiple_of (:obj:`int`, `optional`):
+            If set will pad the sequence to a multiple of the provided value.
+
+            This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >=
+            7.5 (Volta).
+        label_pad_token_id (:obj:`int`, `optional`, defaults to -100):
+            The id to use when padding the labels (-100 will be automatically ignored by PyTorch loss functions).
+    """
+
+    tokenizer: PreTrainedTokenizerBase
+    model: Optional[Any] = None
+    padding: Union[bool, str, PaddingStrategy] = True
+    max_length: Optional[int] = None
+    pad_to_multiple_of: Optional[int] = None
+    label_pad_token_id: int = -100
+    return_tensors: str = "pt"
+    num_labels: int = 0
+    args: Any = None
+
+    def __call__(self, features, return_tensors=None):
+
+        if return_tensors is None:
+            return_tensors = self.return_tensors
+        labels = [feature.pop("labels") for feature in features] if "labels" in features[0].keys() else None
+        label = [feature.pop("label") for feature in features]
+        features_keys = {}
+        name_keys = list(features[0].keys())
+        for k in name_keys:
+            # ignore the padding arguments
+            if k in ["input_ids", "attention_mask", "token_type_ids"]: continue
+            try:
+                features_keys[k] = [feature.pop(k) for feature in features]
+            except KeyError:
+                continue
+
+        # We have to pad the labels before calling `tokenizer.pad` as this method won't pad them and needs them of the
+        # same length to return tensors.
+        bsz = len(labels)
+        with torch.no_grad():
+            new_labels = torch.zeros(bsz, self.num_labels)
+            for i,l in enumerate(labels):
+                if isinstance(l, int): 
+                    new_labels[i][l] = 1
+                else:
+                    if (l[0] != getNegativeEntityId(self.args)):
+                        for j in l:
+                            new_labels[i][j] = 1
+            labels = new_labels
+
+        features = self.tokenizer.pad(
+            features,
+            padding=self.padding,
+            max_length=self.max_length,
+            pad_to_multiple_of=self.pad_to_multiple_of,
+            return_tensors=return_tensors,
+        )
+        features['labels'] = labels
+        features['label'] = torch.tensor(label)
+        features.update(features_keys)
+
+        return features
+
+
+
+class KGC(BaseDataModule):
+    def __init__(self, args, model) -> None:
+        super().__init__(args)
+        self.tokenizer = AutoTokenizer.from_pretrained(self.args.model_name_or_path, use_fast=False)
+        self.processor = KGProcessor(self.tokenizer, args)
+        self.label_list = self.processor.get_labels(args.data_dir)
+
+        entity_list = self.processor.get_entities(args.data_dir)
+        
+        num_added_tokens = self.tokenizer.add_special_tokens({'additional_special_tokens': entity_list})
+        self.sampler = DataCollatorForSeq2Seq(self.tokenizer,
+            model=model,
+            label_pad_token_id=self.tokenizer.pad_token_id,
+            pad_to_multiple_of=8 if self.args.precision == 16 else None,
+            padding="longest",
+            max_length=self.args.max_seq_length,
+            num_labels = len(entity_list),
+            args=args
+        )
+        relations_tokens = self.processor.get_relations(args.data_dir)
+        self.num_relations = len(relations_tokens)
+        num_added_tokens = self.tokenizer.add_special_tokens({'additional_special_tokens': relations_tokens})
+
+        vocab = self.tokenizer.get_added_vocab()
+        self.relation_id_st = vocab[relations_tokens[0]]
+        self.relation_id_ed = vocab[relations_tokens[-1]] + 1
+        self.entity_id_st = vocab[entity_list[0]]
+        self.entity_id_ed = vocab[entity_list[-1]] + 1
+
+
+    def setup(self, stage=None):
+        self.data_train = get_dataset(self.args, self.processor, self.label_list, self.tokenizer, "train")
+        self.data_val = get_dataset(self.args, self.processor, self.label_list, self.tokenizer, "dev")
+        self.data_test = get_dataset(self.args, self.processor, self.label_list, self.tokenizer, "test")
+
+    def prepare_data(self):
+        pass
+
+    def get_config(self):
+        d = {}
+        for k, v in self.__dict__.items():
+            if "st" in k or "ed" in k:
+                d.update({k:v})
+        
+        return d
+
+
+    @staticmethod
+    def add_to_argparse(parser):
+        BaseDataModule.add_to_argparse(parser)
+        parser.add_argument("--model_name_or_path", type=str, default="roberta-base", help="the name or the path to the pretrained model")
+        parser.add_argument("--data_dir", type=str, default="roberta-base", help="the name or the path to the pretrained model")
+        parser.add_argument("--max_seq_length", type=int, default=256, help="Number of examples to operate on per forward step.")
+        parser.add_argument("--warm_up_radio", type=float, default=0.1, help="Number of examples to operate on per forward step.")
+        parser.add_argument("--eval_batch_size", type=int, default=8)
+        parser.add_argument("--overwrite_cache", action="store_true", default=False)
+        return parser
+
+    def get_tokenizer(self):
+        return self.tokenizer
+
+    def train_dataloader(self):
+        return DataLoader(self.data_train, num_workers=self.num_workers, pin_memory=True, collate_fn=self.sampler, batch_size=self.args.batch_size, shuffle=not self.args.faiss_init)
+
+    def val_dataloader(self):
+        return DataLoader(self.data_val, num_workers=self.num_workers, pin_memory=True, collate_fn=self.sampler, batch_size=self.args.eval_batch_size)
+
+    def test_dataloader(self):
+        return DataLoader(self.data_test, num_workers=self.num_workers, pin_memory=True, collate_fn=self.sampler, batch_size=self.args.eval_batch_size)
+

data/icews14/about.txt

@@ -1,15 +0,0 @@
-# triples: 89320 
-# entities: 7128 
-# relations: 12409 
-# timesteps: 208 
-# test triples: 8255 
-# valid triples: 8239 
-# train triples: 72826 
-Measure method:  N/A  
-Target Size :  0  
-Grow Factor:  0  
-Shrink Factor:  0  
-Epsilon Factor: 0  
-Search method: N/A  
-filter_dupes: inter
-nonames: False

data/icews14/time_map.dict

@@ -1,209 +0,0 @@
-0	0	2
-1	3	5
-2	6	7
-3	8	9
-4	10	12
-5	13	14
-6	15	16
-7	17	19
-8	20	21
-9	22	23
-10	24	26
-11	27	28
-12	29	30
-13	31	33
-14	34	35
-15	36	37
-16	38	40
-17	41	42
-18	43	44
-19	45	46
-20	47	48
-21	49	49
-22	50	50
-23	51	51
-24	52	53
-25	54	54
-26	55	55
-27	56	57
-28	58	59
-29	60	61
-30	62	62
-31	63	63
-32	64	65
-33	66	68
-34	69	70
-35	71	71
-36	72	72
-37	73	74
-38	75	76
-39	77	78
-40	79	80
-41	81	82
-42	83	84
-43	85	85
-44	86	87
-45	88	89
-46	90	91
-47	92	93
-48	94	96
-49	97	97
-50	98	99
-51	100	101
-52	102	103
-53	104	105
-54	106	107
-55	108	110
-56	111	112
-57	113	114
-58	115	116
-59	117	118
-60	119	119
-61	120	121
-62	122	124
-63	125	125
-64	126	127
-65	128	129
-66	130	131
-67	132	133
-68	134	135
-69	136	138
-70	139	139
-71	140	140
-72	141	141
-73	142	143
-74	144	145
-75	146	147
-76	148	148
-77	149	150
-78	151	152
-79	153	154
-80	155	155
-81	156	157
-82	158	159
-83	160	161
-84	162	163
-85	164	166
-86	167	167
-87	168	168
-88	169	169
-89	170	170
-90	171	173
-91	174	175
-92	176	177
-93	178	180
-94	181	182
-95	183	183
-96	184	185
-97	186	187
-98	188	188
-99	189	190
-100	191	192
-101	193	194
-102	195	195
-103	196	197
-104	198	199
-105	200	201
-106	202	203
-107	204	205
-108	206	208
-109	209	210
-110	211	212
-111	213	215
-112	216	217
-113	218	219
-114	220	221
-115	222	222
-116	223	224
-117	225	226
-118	227	229
-119	230	231
-120	232	233
-121	234	236
-122	237	238
-123	239	239
-124	240	241
-125	242	243
-126	244	245
-127	246	246
-128	247	248
-129	249	250
-130	251	251
-131	252	252
-132	253	253
-133	254	254
-134	255	256
-135	257	257
-136	258	259
-137	260	261
-138	262	263
-139	264	264
-140	265	265
-141	266	266
-142	267	267
-143	268	269
-144	270	271
-145	272	272
-146	273	273
-147	274	274
-148	275	276
-149	277	278
-150	279	279
-151	280	281
-152	282	283
-153	284	285
-154	286	286
-155	287	287
-156	288	288
-157	289	289
-158	290	291
-159	292	292
-160	293	293
-161	294	294
-162	295	295
-163	296	297
-164	298	299
-165	300	300
-166	301	301
-167	302	303
-168	304	305
-169	306	307
-170	308	309
-171	310	310
-172	311	312
-173	313	313
-174	314	314
-175	315	315
-176	316	316
-177	317	317
-178	318	319
-179	320	320
-180	321	321
-181	322	322
-182	323	323
-183	324	324
-184	325	326
-185	327	327
-186	328	328
-187	329	329
-188	330	330
-189	331	332
-190	333	334
-191	335	335
-192	336	336
-193	337	338
-194	339	340
-195	341	342
-196	343	343
-197	344	344
-198	345	346
-199	347	348
-200	349	349
-201	350	350
-202	351	352
-203	353	355
-204	356	357
-205	358	359
-206	360	362
-207	363	365
-208	366	366

data/icews14_both/about.txt

@@ -1,15 +0,0 @@
-# triples: 86517 
-# entities: 7128 
-# relations: 12409 
-# timesteps: 208 
-# test triples: 8218 
-# valid triples: 8193 
-# train triples: 70106 
-Measure method:  N/A  
-Target Size :  0  
-Grow Factor:  0  
-Shrink Factor:  0  
-Epsilon Factor: 0  
-Search method: N/A  
-filter_dupes: both
-nonames: False

data/icews14_both/time_map.dict

@@ -1,209 +0,0 @@
-0	0	2
-1	3	5
-2	6	7
-3	8	9
-4	10	12
-5	13	14
-6	15	16
-7	17	19
-8	20	21
-9	22	23
-10	24	26
-11	27	28
-12	29	30
-13	31	33
-14	34	35
-15	36	37
-16	38	40
-17	41	42
-18	43	44
-19	45	46
-20	47	48
-21	49	49
-22	50	50
-23	51	51
-24	52	53
-25	54	54
-26	55	55
-27	56	57
-28	58	59
-29	60	61
-30	62	62
-31	63	63
-32	64	65
-33	66	68
-34	69	70
-35	71	71
-36	72	72
-37	73	74
-38	75	76
-39	77	78
-40	79	80
-41	81	82
-42	83	84
-43	85	85
-44	86	87
-45	88	89
-46	90	91
-47	92	93
-48	94	96
-49	97	97
-50	98	99
-51	100	101
-52	102	103
-53	104	105
-54	106	107
-55	108	110
-56	111	112
-57	113	114
-58	115	116
-59	117	118
-60	119	119
-61	120	121
-62	122	124
-63	125	125
-64	126	127
-65	128	129
-66	130	131
-67	132	133
-68	134	135
-69	136	138
-70	139	139
-71	140	140
-72	141	141
-73	142	143
-74	144	145
-75	146	147
-76	148	148
-77	149	150
-78	151	152
-79	153	154
-80	155	155
-81	156	157
-82	158	159
-83	160	161
-84	162	163
-85	164	166
-86	167	167
-87	168	168
-88	169	169
-89	170	170
-90	171	173
-91	174	175
-92	176	177
-93	178	180
-94	181	182
-95	183	183
-96	184	185
-97	186	187
-98	188	188
-99	189	190
-100	191	192
-101	193	194
-102	195	195
-103	196	197
-104	198	199
-105	200	201
-106	202	203
-107	204	205
-108	206	208
-109	209	210
-110	211	212
-111	213	215
-112	216	217
-113	218	219
-114	220	221
-115	222	222
-116	223	224
-117	225	226
-118	227	229
-119	230	231
-120	232	233
-121	234	236
-122	237	238
-123	239	239
-124	240	241
-125	242	243
-126	244	245
-127	246	246
-128	247	248
-129	249	250
-130	251	251
-131	252	252
-132	253	253
-133	254	254
-134	255	256
-135	257	257
-136	258	259
-137	260	261
-138	262	263
-139	264	264
-140	265	265
-141	266	266
-142	267	267
-143	268	269
-144	270	271
-145	272	272
-146	273	273
-147	274	274
-148	275	276
-149	277	278
-150	279	279
-151	280	281
-152	282	283
-153	284	285
-154	286	286
-155	287	287
-156	288	288
-157	289	289
-158	290	291
-159	292	292
-160	293	293
-161	294	294
-162	295	295
-163	296	297
-164	298	299
-165	300	300
-166	301	301
-167	302	303
-168	304	305
-169	306	307
-170	308	309
-171	310	310
-172	311	312
-173	313	313
-174	314	314
-175	315	315
-176	316	316
-177	317	317
-178	318	319
-179	320	320
-180	321	321
-181	322	322
-182	323	323
-183	324	324
-184	325	326
-185	327	327
-186	328	328
-187	329	329
-188	330	330
-189	331	332
-190	333	334
-191	335	335
-192	336	336
-193	337	338
-194	339	340
-195	341	342
-196	343	343
-197	344	344
-198	345	346
-199	347	348
-200	349	349
-201	350	350
-202	351	352
-203	353	355
-204	356	357
-205	358	359
-206	360	362
-207	363	365
-208	366	366

data/wikidata12k/about.txt

@@ -1,15 +0,0 @@
-# triples: 291818 
-# entities: 12554 
-# relations: 423 
-# timesteps: 70 
-# test triples: 19271 
-# valid triples: 20208 
-# train triples: 252339 
-Measure method:  N/A  
-Target Size :  423  
-Grow Factor:  0  
-Shrink Factor:  4.0  
-Epsilon Factor: 0  
-Search method: N/A  
-filter_dupes: inter
-nonames: False

data/wikidata12k/relations.dict

@@ -1,423 +0,0 @@
-0	P131[0-0]
-1	P131[1-1]
-2	P131[2-2]
-3	P131[3-3]
-4	P131[4-4]
-5	P131[5-5]
-6	P131[6-6]
-7	P131[7-7]
-8	P131[8-8]
-9	P131[9-9]
-10	P131[10-10]
-11	P131[11-11]
-12	P131[12-12]
-13	P131[13-13]
-14	P131[14-14]
-15	P131[15-15]
-16	P131[16-16]
-17	P131[17-17]
-18	P131[18-18]
-19	P131[19-19]
-20	P131[20-20]
-21	P131[21-21]
-22	P131[22-22]
-23	P131[23-23]
-24	P131[24-24]
-25	P131[25-25]
-26	P131[26-26]
-27	P131[27-27]
-28	P131[28-28]
-29	P131[29-29]
-30	P131[30-30]
-31	P131[31-31]
-32	P131[32-32]
-33	P131[33-33]
-34	P131[34-34]
-35	P131[35-35]
-36	P131[36-36]
-37	P131[37-37]
-38	P131[38-38]
-39	P131[39-39]
-40	P131[40-40]
-41	P131[41-41]
-42	P131[42-42]
-43	P131[43-43]
-44	P131[44-44]
-45	P131[45-45]
-46	P131[46-46]
-47	P131[47-47]
-48	P131[48-48]
-49	P131[49-49]
-50	P131[50-50]
-51	P131[51-51]
-52	P131[52-52]
-53	P131[53-53]
-54	P131[54-54]
-55	P131[55-55]
-56	P131[56-56]
-57	P131[57-57]
-58	P131[58-58]
-59	P131[59-59]
-60	P131[60-60]
-61	P131[61-61]
-62	P131[62-62]
-63	P131[63-63]
-64	P131[64-64]
-65	P131[65-65]
-66	P131[66-66]
-67	P131[67-67]
-68	P131[68-68]
-69	P131[69-69]
-70	P1435[65-65]
-71	P39[49-49]
-72	P39[50-50]
-73	P39[51-51]
-74	P39[52-52]
-75	P39[53-53]
-76	P39[54-54]
-77	P39[55-55]
-78	P39[56-56]
-79	P39[57-57]
-80	P39[58-58]
-81	P39[59-59]
-82	P39[60-60]
-83	P39[61-61]
-84	P39[62-62]
-85	P39[63-63]
-86	P39[64-64]
-87	P39[65-65]
-88	P39[66-66]
-89	P39[67-67]
-90	P39[68-68]
-91	P39[69-69]
-92	P54[40-40]
-93	P54[41-41]
-94	P54[42-42]
-95	P54[43-43]
-96	P54[44-44]
-97	P54[45-45]
-98	P54[46-46]
-99	P54[47-47]
-100	P54[48-48]
-101	P54[49-49]
-102	P54[50-50]
-103	P54[51-51]
-104	P54[52-52]
-105	P54[53-53]
-106	P54[54-54]
-107	P54[55-55]
-108	P54[56-56]
-109	P54[57-57]
-110	P54[58-58]
-111	P54[59-59]
-112	P54[60-60]
-113	P54[61-61]
-114	P54[62-62]
-115	P54[63-63]
-116	P54[64-64]
-117	P54[65-65]
-118	P54[66-66]
-119	P54[67-67]
-120	P54[68-68]
-121	P54[69-69]
-122	P31[0-0]
-123	P31[1-1]
-124	P31[2-2]
-125	P31[3-3]
-126	P31[4-4]
-127	P31[5-5]
-128	P31[6-6]
-129	P31[7-7]
-130	P31[8-8]
-131	P31[9-9]
-132	P31[10-10]
-133	P31[11-11]
-134	P31[12-12]
-135	P31[13-13]
-136	P31[14-14]
-137	P31[15-15]
-138	P31[16-16]
-139	P31[17-17]
-140	P31[18-18]
-141	P31[19-19]
-142	P31[20-20]
-143	P31[21-21]
-144	P31[22-22]
-145	P31[23-23]
-146	P31[24-24]
-147	P31[25-25]
-148	P31[26-26]
-149	P31[27-27]
-150	P31[28-28]
-151	P31[29-29]
-152	P31[30-30]
-153	P31[31-31]
-154	P31[32-32]
-155	P31[33-33]
-156	P31[34-34]
-157	P31[35-35]
-158	P31[36-36]
-159	P31[37-37]
-160	P31[38-38]
-161	P31[39-39]
-162	P31[40-40]
-163	P31[41-41]
-164	P31[42-42]
-165	P31[43-43]
-166	P31[44-44]
-167	P31[45-45]
-168	P31[46-46]
-169	P31[47-47]
-170	P31[48-48]
-171	P31[49-49]
-172	P31[50-50]
-173	P31[51-51]
-174	P31[52-52]
-175	P31[53-53]
-176	P31[54-54]
-177	P31[55-55]
-178	P31[56-56]
-179	P31[57-57]
-180	P31[58-58]
-181	P31[59-59]
-182	P31[60-60]
-183	P31[61-61]
-184	P31[62-62]
-185	P31[63-63]
-186	P31[64-64]
-187	P31[65-65]
-188	P31[66-66]
-189	P31[67-67]
-190	P31[68-68]
-191	P31[69-69]
-192	P463[26-26]
-193	P463[27-27]
-194	P463[28-28]
-195	P463[29-29]
-196	P463[30-30]
-197	P463[31-31]
-198	P463[32-32]
-199	P463[33-33]
-200	P463[34-34]
-201	P463[35-35]
-202	P463[36-36]
-203	P463[37-37]
-204	P463[38-38]
-205	P463[39-39]
-206	P463[40-40]
-207	P463[41-41]
-208	P463[42-42]
-209	P463[43-43]
-210	P463[44-44]
-211	P463[45-45]
-212	P463[46-46]
-213	P463[47-47]
-214	P463[48-48]
-215	P463[49-49]
-216	P463[50-50]
-217	P463[51-51]
-218	P463[52-52]
-219	P463[53-53]
-220	P463[54-54]
-221	P463[55-55]
-222	P463[56-56]
-223	P463[57-57]
-224	P463[58-58]
-225	P463[59-59]
-226	P463[60-60]
-227	P463[61-61]
-228	P463[62-62]
-229	P463[63-63]
-230	P463[64-64]
-231	P463[65-65]
-232	P463[66-66]
-233	P463[67-67]
-234	P463[68-68]
-235	P463[69-69]
-236	P512[4-69]
-237	P190[0-29]
-238	P150[0-3]
-239	P1376[39-47]
-240	P463[0-7]
-241	P166[0-7]
-242	P2962[18-30]
-243	P108[29-36]
-244	P39[0-3]
-245	P17[47-48]
-246	P166[21-23]
-247	P793[46-69]
-248	P69[32-41]
-249	P17[57-58]
-250	P190[42-45]
-251	P2962[39-42]
-252	P54[0-18]
-253	P26[56-61]
-254	P150[14-17]
-255	P463[16-17]
-256	P26[39-46]
-257	P579[36-43]
-258	P579[16-23]
-259	P2962[59-60]
-260	P1411[59-61]
-261	P26[20-27]
-262	P6[4-69]
-263	P1435[33-34]
-264	P166[52-53]
-265	P108[49-57]
-266	P150[10-13]
-267	P1346[47-68]
-268	P150[18-21]
-269	P1346[13-46]
-270	P69[20-23]
-271	P39[31-32]
-272	P1411[32-37]
-273	P166[62-63]
-274	P150[44-47]
-275	P2962[61-62]
-276	P150[48-51]
-277	P150[52-55]
-278	P1411[62-67]
-279	P1435[35-36]
-280	P1411[48-51]
-281	P150[22-25]
-282	P2962[63-64]
-283	P2962[65-66]
-284	P166[58-59]
-285	P190[46-49]
-286	P54[34-35]
-287	P1435[4-16]
-288	P463[18-19]
-289	P150[31-34]
-290	P150[35-38]
-291	P39[35-36]
-292	P26[62-69]
-293	P1411[56-58]
-294	P1435[37-38]
-295	P166[60-61]
-296	P39[33-34]
-297	P102[24-31]
-298	P2962[43-46]
-299	P108[37-48]
-300	P190[50-53]
-301	P39[4-6]
-302	P1435[39-40]
-303	P793[0-45]
-304	P150[64-69]
-305	P39[19-22]
-306	P27[30-38]
-307	P2962[31-38]
-308	P1411[24-31]
-309	P102[40-45]
-310	P39[37-38]
-311	P463[8-11]
-312	P1435[41-42]
-313	P27[52-59]
-314	P69[16-19]
-315	P17[16-18]
-316	P190[54-57]
-317	P1435[43-44]
-318	P166[8-15]
-319	P166[45-47]
-320	P2962[47-50]
-321	P39[39-40]
-322	P1411[52-55]
-323	P108[58-69]
-324	P463[20-21]
-325	P39[41-42]
-326	P150[26-30]
-327	P150[39-43]
-328	P1435[45-46]
-329	P26[28-38]
-330	P54[27-30]
-331	P190[58-61]
-332	P17[59-61]
-333	P54[36-37]
-334	P166[16-20]
-335	P166[37-40]
-336	P1435[47-48]
-337	P17[0-3]
-338	P26[47-55]
-339	P1435[49-50]
-340	P1435[25-28]
-341	P150[4-9]
-342	P102[63-69]
-343	P26[0-19]
-344	P1435[17-24]
-345	P39[23-26]
-346	P1435[51-52]
-347	P39[7-11]
-348	P69[12-15]
-349	P69[24-31]
-350	P102[0-23]
-351	P39[43-44]
-352	P579[24-35]
-353	P190[62-65]
-354	P1435[53-54]
-355	P1376[0-18]
-356	P27[0-14]
-357	P463[12-15]
-358	P166[33-36]
-359	P102[32-39]
-360	P17[4-7]
-361	P190[30-41]
-362	P166[24-28]
-363	P190[66-69]
-364	P69[42-69]
-365	P1435[55-56]
-366	P54[31-33]
-367	P39[45-46]
-368	P17[12-15]
-369	P1435[57-58]
-370	P54[19-26]
-371	P2962[51-54]
-372	P2962[67-69]
-373	P1435[59-60]
-374	P579[44-56]
-375	P1435[61-62]
-376	P166[41-44]
-377	P17[19-22]
-378	P1376[19-38]
-379	P17[23-26]
-380	P1376[48-69]
-381	P463[22-23]
-382	P17[27-30]
-383	P1435[63-64]
-384	P69[0-3]
-385	P1435[66-67]
-386	P17[35-38]
-387	P69[8-11]
-388	P1435[68-69]
-389	P17[31-34]
-390	P102[46-53]
-391	P27[60-69]
-392	P579[57-69]
-393	P69[4-7]
-394	P1411[7-14]
-395	P551[0-35]
-396	P108[0-28]
-397	P17[8-11]
-398	P1411[38-47]
-399	P17[43-46]
-400	P17[49-52]
-401	P166[64-69]
-402	P1435[29-32]
-403	P54[38-39]
-404	P39[27-30]
-405	P2962[55-58]
-406	P463[24-25]
-407	P17[39-42]
-408	P17[53-56]
-409	P17[66-69]
-410	P17[62-65]
-411	P1411[15-23]
-412	P166[48-51]
-413	P27[15-29]
-414	P150[56-63]
-415	P27[39-51]
-416	P39[47-48]
-417	P166[29-32]
-418	P39[12-18]
-419	P166[54-57]
-420	P551[36-69]
-421	P579[0-15]
-422	P102[54-62]

data/wikidata12k/time_map.dict

@@ -1,71 +0,0 @@
-0	19	19
-1	20	1643
-2	1644	1790
-3	1791	1816
-4	1817	1855
-5	1856	1871
-6	1872	1893
-7	1894	1905
-8	1906	1913
-9	1914	1918
-10	1919	1920
-11	1921	1924
-12	1925	1929
-13	1930	1933
-14	1934	1937
-15	1938	1941
-16	1942	1945
-17	1946	1948
-18	1949	1950
-19	1951	1953
-20	1954	1956
-21	1957	1959
-22	1960	1961
-23	1962	1963
-24	1964	1965
-25	1966	1967
-26	1968	1968
-27	1969	1970
-28	1971	1972
-29	1973	1974
-30	1975	1976
-31	1977	1978
-32	1979	1980
-33	1981	1982
-34	1983	1983
-35	1984	1984
-36	1985	1985
-37	1986	1986
-38	1987	1987
-39	1988	1988
-40	1989	1989
-41	1990	1990
-42	1991	1991
-43	1992	1992
-44	1993	1993
-45	1994	1994
-46	1995	1995
-47	1996	1996
-48	1997	1997
-49	1998	1998
-50	1999	1999
-51	2000	2000
-52	2001	2001
-53	2002	2002
-54	2003	2003
-55	2004	2004
-56	2005	2005
-57	2006	2006
-58	2007	2007
-59	2008	2008
-60	2009	2009
-61	2010	2010
-62	2011	2011
-63	2012	2012
-64	2013	2013
-65	2014	2014
-66	2015	2015
-67	2016	2016
-68	2017	2017
-69	2018	2020
-70	2021	2021

data/wikidata12k_both/about.txt

@@ -1,15 +0,0 @@
-# triples: 231529 
-# entities: 12554 
-# relations: 423 
-# timesteps: 70 
-# test triples: 16195 
-# valid triples: 16707 
-# train triples: 198627 
-Measure method:  N/A  
-Target Size :  423  
-Grow Factor:  0  
-Shrink Factor:  4.0  
-Epsilon Factor: 0  
-Search method: N/A  
-filter_dupes: both
-nonames: False

data/wikidata12k_both/relations.dict

@@ -1,423 +0,0 @@
-0	P131[0-0]
-1	P131[1-1]
-2	P131[2-2]
-3	P131[3-3]
-4	P131[4-4]
-5	P131[5-5]
-6	P131[6-6]
-7	P131[7-7]
-8	P131[8-8]
-9	P131[9-9]
-10	P131[10-10]
-11	P131[11-11]
-12	P131[12-12]
-13	P131[13-13]
-14	P131[14-14]
-15	P131[15-15]
-16	P131[16-16]
-17	P131[17-17]
-18	P131[18-18]
-19	P131[19-19]
-20	P131[20-20]
-21	P131[21-21]
-22	P131[22-22]
-23	P131[23-23]
-24	P131[24-24]
-25	P131[25-25]
-26	P131[26-26]
-27	P131[27-27]
-28	P131[28-28]
-29	P131[29-29]
-30	P131[30-30]
-31	P131[31-31]
-32	P131[32-32]
-33	P131[33-33]
-34	P131[34-34]
-35	P131[35-35]
-36	P131[36-36]
-37	P131[37-37]
-38	P131[38-38]
-39	P131[39-39]
-40	P131[40-40]
-41	P131[41-41]
-42	P131[42-42]
-43	P131[43-43]
-44	P131[44-44]
-45	P131[45-45]
-46	P131[46-46]
-47	P131[47-47]
-48	P131[48-48]
-49	P131[49-49]
-50	P131[50-50]
-51	P131[51-51]
-52	P131[52-52]
-53	P131[53-53]
-54	P131[54-54]
-55	P131[55-55]
-56	P131[56-56]
-57	P131[57-57]
-58	P131[58-58]
-59	P131[59-59]
-60	P131[60-60]
-61	P131[61-61]
-62	P131[62-62]
-63	P131[63-63]
-64	P131[64-64]
-65	P131[65-65]
-66	P131[66-66]
-67	P131[67-67]
-68	P131[68-68]
-69	P131[69-69]
-70	P1435[65-65]
-71	P39[49-49]
-72	P39[50-50]
-73	P39[51-51]
-74	P39[52-52]
-75	P39[53-53]
-76	P39[54-54]
-77	P39[55-55]
-78	P39[56-56]
-79	P39[57-57]
-80	P39[58-58]
-81	P39[59-59]
-82	P39[60-60]
-83	P39[61-61]
-84	P39[62-62]
-85	P39[63-63]
-86	P39[64-64]
-87	P39[65-65]
-88	P39[66-66]
-89	P39[67-67]
-90	P39[68-68]
-91	P39[69-69]
-92	P54[40-40]
-93	P54[41-41]
-94	P54[42-42]
-95	P54[43-43]
-96	P54[44-44]
-97	P54[45-45]
-98	P54[46-46]
-99	P54[47-47]
-100	P54[48-48]
-101	P54[49-49]
-102	P54[50-50]
-103	P54[51-51]
-104	P54[52-52]
-105	P54[53-53]
-106	P54[54-54]
-107	P54[55-55]
-108	P54[56-56]
-109	P54[57-57]
-110	P54[58-58]
-111	P54[59-59]
-112	P54[60-60]
-113	P54[61-61]
-114	P54[62-62]
-115	P54[63-63]
-116	P54[64-64]
-117	P54[65-65]
-118	P54[66-66]
-119	P54[67-67]
-120	P54[68-68]
-121	P54[69-69]
-122	P31[0-0]
-123	P31[1-1]
-124	P31[2-2]
-125	P31[3-3]
-126	P31[4-4]
-127	P31[5-5]
-128	P31[6-6]
-129	P31[7-7]
-130	P31[8-8]
-131	P31[9-9]
-132	P31[10-10]
-133	P31[11-11]
-134	P31[12-12]
-135	P31[13-13]
-136	P31[14-14]
-137	P31[15-15]
-138	P31[16-16]
-139	P31[17-17]
-140	P31[18-18]
-141	P31[19-19]
-142	P31[20-20]
-143	P31[21-21]
-144	P31[22-22]
-145	P31[23-23]
-146	P31[24-24]
-147	P31[25-25]
-148	P31[26-26]
-149	P31[27-27]
-150	P31[28-28]
-151	P31[29-29]
-152	P31[30-30]
-153	P31[31-31]
-154	P31[32-32]
-155	P31[33-33]
-156	P31[34-34]
-157	P31[35-35]
-158	P31[36-36]
-159	P31[37-37]
-160	P31[38-38]
-161	P31[39-39]
-162	P31[40-40]
-163	P31[41-41]
-164	P31[42-42]
-165	P31[43-43]
-166	P31[44-44]
-167	P31[45-45]
-168	P31[46-46]
-169	P31[47-47]
-170	P31[48-48]
-171	P31[49-49]
-172	P31[50-50]
-173	P31[51-51]
-174	P31[52-52]
-175	P31[53-53]
-176	P31[54-54]
-177	P31[55-55]
-178	P31[56-56]
-179	P31[57-57]
-180	P31[58-58]
-181	P31[59-59]
-182	P31[60-60]
-183	P31[61-61]
-184	P31[62-62]
-185	P31[63-63]
-186	P31[64-64]
-187	P31[65-65]
-188	P31[66-66]
-189	P31[67-67]
-190	P31[68-68]
-191	P31[69-69]
-192	P463[26-26]
-193	P463[27-27]
-194	P463[28-28]
-195	P463[29-29]
-196	P463[30-30]
-197	P463[31-31]
-198	P463[32-32]
-199	P463[33-33]
-200	P463[34-34]
-201	P463[35-35]
-202	P463[36-36]
-203	P463[37-37]
-204	P463[38-38]
-205	P463[39-39]
-206	P463[40-40]
-207	P463[41-41]
-208	P463[42-42]
-209	P463[43-43]
-210	P463[44-44]
-211	P463[45-45]
-212	P463[46-46]
-213	P463[47-47]
-214	P463[48-48]
-215	P463[49-49]
-216	P463[50-50]
-217	P463[51-51]
-218	P463[52-52]
-219	P463[53-53]
-220	P463[54-54]
-221	P463[55-55]
-222	P463[56-56]
-223	P463[57-57]
-224	P463[58-58]
-225	P463[59-59]
-226	P463[60-60]
-227	P463[61-61]
-228	P463[62-62]
-229	P463[63-63]
-230	P463[64-64]
-231	P463[65-65]
-232	P463[66-66]
-233	P463[67-67]
-234	P463[68-68]
-235	P463[69-69]
-236	P512[4-69]
-237	P190[0-29]
-238	P150[0-3]
-239	P1376[39-47]
-240	P463[0-7]
-241	P166[0-7]
-242	P2962[18-30]
-243	P108[29-36]
-244	P39[0-3]
-245	P17[47-48]
-246	P166[21-23]
-247	P793[46-69]
-248	P69[32-41]
-249	P17[57-58]
-250	P190[42-45]
-251	P2962[39-42]
-252	P54[0-18]
-253	P26[56-61]
-254	P150[14-17]
-255	P463[16-17]
-256	P26[39-46]
-257	P579[36-43]
-258	P579[16-23]
-259	P2962[59-60]
-260	P1411[59-61]
-261	P26[20-27]
-262	P6[4-69]
-263	P1435[33-34]
-264	P166[52-53]
-265	P108[49-57]
-266	P150[10-13]
-267	P1346[47-68]
-268	P150[18-21]
-269	P1346[13-46]
-270	P69[20-23]
-271	P39[31-32]
-272	P1411[32-37]
-273	P166[62-63]
-274	P150[44-47]
-275	P2962[61-62]
-276	P150[48-51]
-277	P150[52-55]
-278	P1411[62-67]
-279	P1435[35-36]
-280	P1411[48-51]
-281	P150[22-25]
-282	P2962[63-64]
-283	P2962[65-66]
-284	P166[58-59]
-285	P190[46-49]
-286	P54[34-35]
-287	P1435[4-16]
-288	P463[18-19]
-289	P150[31-34]
-290	P150[35-38]
-291	P39[35-36]
-292	P26[62-69]
-293	P1411[56-58]
-294	P1435[37-38]
-295	P166[60-61]
-296	P39[33-34]
-297	P102[24-31]
-298	P2962[43-46]
-299	P108[37-48]
-300	P190[50-53]
-301	P39[4-6]
-302	P1435[39-40]
-303	P793[0-45]
-304	P150[64-69]
-305	P39[19-22]
-306	P27[30-38]
-307	P2962[31-38]
-308	P1411[24-31]
-309	P102[40-45]
-310	P39[37-38]
-311	P463[8-11]
-312	P1435[41-42]
-313	P27[52-59]
-314	P69[16-19]
-315	P17[16-18]
-316	P190[54-57]
-317	P1435[43-44]
-318	P166[8-15]
-319	P166[45-47]
-320	P2962[47-50]
-321	P39[39-40]
-322	P1411[52-55]
-323	P108[58-69]
-324	P463[20-21]
-325	P39[41-42]
-326	P150[26-30]
-327	P150[39-43]
-328	P1435[45-46]
-329	P26[28-38]
-330	P54[27-30]
-331	P190[58-61]
-332	P17[59-61]
-333	P54[36-37]
-334	P166[16-20]
-335	P166[37-40]
-336	P1435[47-48]
-337	P17[0-3]
-338	P26[47-55]
-339	P1435[49-50]
-340	P1435[25-28]
-341	P150[4-9]
-342	P102[63-69]
-343	P26[0-19]
-344	P1435[17-24]
-345	P39[23-26]
-346	P1435[51-52]
-347	P39[7-11]
-348	P69[12-15]
-349	P69[24-31]
-350	P102[0-23]
-351	P39[43-44]
-352	P579[24-35]
-353	P190[62-65]
-354	P1435[53-54]
-355	P1376[0-18]
-356	P27[0-14]
-357	P463[12-15]
-358	P166[33-36]
-359	P102[32-39]
-360	P17[4-7]
-361	P190[30-41]
-362	P166[24-28]
-363	P190[66-69]
-364	P69[42-69]
-365	P1435[55-56]
-366	P54[31-33]
-367	P39[45-46]
-368	P17[12-15]
-369	P1435[57-58]
-370	P54[19-26]
-371	P2962[51-54]
-372	P2962[67-69]
-373	P1435[59-60]
-374	P579[44-56]
-375	P1435[61-62]
-376	P166[41-44]
-377	P17[19-22]
-378	P1376[19-38]
-379	P17[23-26]
-380	P1376[48-69]
-381	P463[22-23]
-382	P17[27-30]
-383	P1435[63-64]
-384	P69[0-3]
-385	P1435[66-67]
-386	P17[35-38]
-387	P69[8-11]
-388	P1435[68-69]
-389	P17[31-34]
-390	P102[46-53]
-391	P27[60-69]
-392	P579[57-69]
-393	P69[4-7]
-394	P1411[7-14]
-395	P551[0-35]
-396	P108[0-28]
-397	P17[8-11]
-398	P1411[38-47]
-399	P17[43-46]
-400	P17[49-52]
-401	P166[64-69]
-402	P1435[29-32]
-403	P54[38-39]
-404	P39[27-30]
-405	P2962[55-58]
-406	P463[24-25]
-407	P17[39-42]
-408	P17[53-56]
-409	P17[66-69]
-410	P17[62-65]
-411	P1411[15-23]
-412	P166[48-51]
-413	P27[15-29]
-414	P150[56-63]
-415	P27[39-51]
-416	P39[47-48]
-417	P166[29-32]
-418	P39[12-18]
-419	P166[54-57]
-420	P551[36-69]
-421	P579[0-15]
-422	P102[54-62]

data/wikidata12k_both/time_map.dict

@@ -1,71 +0,0 @@
-0	19	19
-1	20	1643
-2	1644	1790
-3	1791	1816
-4	1817	1855
-5	1856	1871
-6	1872	1893
-7	1894	1905
-8	1906	1913
-9	1914	1918
-10	1919	1920
-11	1921	1924
-12	1925	1929
-13	1930	1933
-14	1934	1937
-15	1938	1941
-16	1942	1945
-17	1946	1948
-18	1949	1950
-19	1951	1953
-20	1954	1956
-21	1957	1959
-22	1960	1961
-23	1962	1963
-24	1964	1965
-25	1966	1967
-26	1968	1968
-27	1969	1970
-28	1971	1972
-29	1973	1974
-30	1975	1976
-31	1977	1978
-32	1979	1980
-33	1981	1982
-34	1983	1983
-35	1984	1984
-36	1985	1985
-37	1986	1986
-38	1987	1987
-39	1988	1988
-40	1989	1989
-41	1990	1990
-42	1991	1991
-43	1992	1992
-44	1993	1993
-45	1994	1994
-46	1995	1995
-47	1996	1996
-48	1997	1997
-49	1998	1998
-50	1999	1999
-51	2000	2000
-52	2001	2001
-53	2002	2002
-54	2003	2003
-55	2004	2004
-56	2005	2005
-57	2006	2006
-58	2007	2007
-59	2008	2008
-60	2009	2009
-61	2010	2010
-62	2011	2011
-63	2012	2012
-64	2013	2013
-65	2014	2014
-66	2015	2015
-67	2016	2016
-68	2017	2017
-69	2018	2020
-70	2021	2021

data/yago11k/about.txt

@@ -1,15 +0,0 @@
-# triples: 78032 
-# entities: 10526 
-# relations: 177 
-# timesteps: 46 
-# test triples: 6909 
-# valid triples: 7198 
-# train triples: 63925 
-Measure method:  N/A  
-Target Size :  0  
-Grow Factor:  0  
-Shrink Factor:  0  
-Epsilon Factor: 5.0  
-Search method: N/A  
-filter_dupes: inter
-nonames: False

data/yago11k/relations.dict

@@ -1,177 +0,0 @@
-0	<wasBornIn>[0-2]
-1	<wasBornIn>[2-5]
-2	<wasBornIn>[5-7]
-3	<wasBornIn>[7-10]
-4	<wasBornIn>[10-12]
-5	<wasBornIn>[12-15]
-6	<wasBornIn>[15-17]
-7	<wasBornIn>[17-20]
-8	<wasBornIn>[20-22]
-9	<wasBornIn>[22-25]
-10	<wasBornIn>[25-27]
-11	<wasBornIn>[27-30]
-12	<wasBornIn>[30-32]
-13	<wasBornIn>[32-35]
-14	<wasBornIn>[35-45]
-15	<wasBornIn>[52-52]
-16	<diedIn>[0-3]
-17	<diedIn>[3-5]
-18	<diedIn>[5-7]
-19	<diedIn>[7-10]
-20	<diedIn>[10-12]
-21	<diedIn>[12-14]
-22	<diedIn>[14-17]
-23	<diedIn>[17-19]
-24	<diedIn>[19-21]
-25	<diedIn>[21-23]
-26	<diedIn>[23-25]
-27	<diedIn>[25-27]
-28	<diedIn>[27-29]
-29	<diedIn>[29-32]
-30	<diedIn>[32-34]
-31	<diedIn>[34-36]
-32	<diedIn>[36-38]
-33	<diedIn>[38-40]
-34	<diedIn>[40-42]
-35	<diedIn>[42-44]
-36	<diedIn>[44-47]
-37	<diedIn>[47-49]
-38	<diedIn>[49-51]
-39	<diedIn>[51-53]
-40	<diedIn>[53-55]
-41	<diedIn>[55-57]
-42	<diedIn>[59-59]
-43	<worksAt>[0-3]
-44	<worksAt>[3-5]
-45	<worksAt>[5-7]
-46	<worksAt>[7-10]
-47	<worksAt>[10-12]
-48	<worksAt>[12-14]
-49	<worksAt>[14-17]
-50	<worksAt>[17-19]
-51	<worksAt>[19-21]
-52	<worksAt>[21-23]
-53	<worksAt>[23-25]
-54	<worksAt>[25-27]
-55	<worksAt>[27-29]
-56	<worksAt>[29-32]
-57	<worksAt>[32-34]
-58	<worksAt>[34-36]
-59	<worksAt>[36-40]
-60	<worksAt>[40-42]
-61	<worksAt>[42-47]
-62	<worksAt>[47-53]
-63	<worksAt>[59-59]
-64	<playsFor>[0-3]
-65	<playsFor>[3-5]
-66	<playsFor>[5-23]
-67	<playsFor>[23-25]
-68	<playsFor>[25-27]
-69	<playsFor>[27-29]
-70	<playsFor>[29-32]
-71	<playsFor>[32-34]
-72	<playsFor>[34-36]
-73	<playsFor>[36-38]
-74	<playsFor>[38-40]
-75	<playsFor>[40-42]
-76	<playsFor>[42-44]
-77	<playsFor>[44-47]
-78	<playsFor>[47-51]
-79	<playsFor>[59-59]
-80	<hasWonPrize>[1-4]
-81	<hasWonPrize>[4-6]
-82	<hasWonPrize>[6-8]
-83	<hasWonPrize>[8-11]
-84	<hasWonPrize>[11-15]
-85	<hasWonPrize>[15-18]
-86	<hasWonPrize>[18-22]
-87	<hasWonPrize>[22-26]
-88	<hasWonPrize>[26-30]
-89	<hasWonPrize>[30-33]
-90	<hasWonPrize>[33-37]
-91	<hasWonPrize>[37-47]
-92	<hasWonPrize>[47-53]
-93	<hasWonPrize>[59-59]
-94	<isMarriedTo>[0-3]
-95	<isMarriedTo>[3-5]
-96	<isMarriedTo>[5-7]
-97	<isMarriedTo>[7-10]
-98	<isMarriedTo>[10-12]
-99	<isMarriedTo>[12-14]
-100	<isMarriedTo>[14-17]
-101	<isMarriedTo>[17-19]
-102	<isMarriedTo>[19-21]
-103	<isMarriedTo>[21-23]
-104	<isMarriedTo>[23-25]
-105	<isMarriedTo>[25-27]
-106	<isMarriedTo>[27-29]
-107	<isMarriedTo>[29-32]
-108	<isMarriedTo>[32-34]
-109	<isMarriedTo>[34-38]
-110	<isMarriedTo>[38-42]
-111	<isMarriedTo>[42-47]
-112	<isMarriedTo>[47-51]
-113	<isMarriedTo>[51-55]
-114	<isMarriedTo>[59-59]
-115	<owns>[0-10]
-116	<owns>[10-17]
-117	<owns>[17-19]
-118	<owns>[19-23]
-119	<owns>[23-36]
-120	<owns>[36-38]
-121	<owns>[59-59]
-122	<graduatedFrom>[0-3]
-123	<graduatedFrom>[3-5]
-124	<graduatedFrom>[5-7]
-125	<graduatedFrom>[7-10]
-126	<graduatedFrom>[10-14]
-127	<graduatedFrom>[14-17]
-128	<graduatedFrom>[17-19]
-129	<graduatedFrom>[19-21]
-130	<graduatedFrom>[21-23]
-131	<graduatedFrom>[23-27]
-132	<graduatedFrom>[27-32]
-133	<graduatedFrom>[32-34]
-134	<graduatedFrom>[34-38]
-135	<graduatedFrom>[38-42]
-136	<graduatedFrom>[59-59]
-137	<isAffiliatedTo>[1-4]
-138	<isAffiliatedTo>[4-6]
-139	<isAffiliatedTo>[6-8]
-140	<isAffiliatedTo>[8-11]
-141	<isAffiliatedTo>[11-13]
-142	<isAffiliatedTo>[13-15]
-143	<isAffiliatedTo>[15-18]
-144	<isAffiliatedTo>[18-20]
-145	<isAffiliatedTo>[20-22]
-146	<isAffiliatedTo>[22-24]
-147	<isAffiliatedTo>[24-26]
-148	<isAffiliatedTo>[26-28]
-149	<isAffiliatedTo>[28-30]
-150	<isAffiliatedTo>[30-33]
-151	<isAffiliatedTo>[33-35]
-152	<isAffiliatedTo>[35-37]
-153	<isAffiliatedTo>[37-40]
-154	<isAffiliatedTo>[40-42]
-155	<isAffiliatedTo>[42-44]
-156	<isAffiliatedTo>[44-47]
-157	<isAffiliatedTo>[47-49]
-158	<isAffiliatedTo>[49-51]
-159	<isAffiliatedTo>[51-53]
-160	<isAffiliatedTo>[53-55]
-161	<isAffiliatedTo>[55-57]
-162	<isAffiliatedTo>[59-59]
-163	<created>[0-3]
-164	<created>[3-5]
-165	<created>[5-10]
-166	<created>[10-12]
-167	<created>[12-17]
-168	<created>[17-19]
-169	<created>[19-25]
-170	<created>[25-29]
-171	<created>[29-32]
-172	<created>[32-36]
-173	<created>[36-42]
-174	<created>[42-47]
-175	<created>[47-53]
-176	<created>[59-59]

data/yago11k/time_map.dict

@@ -1,60 +0,0 @@
-0	-431	1782
-1	1783	1848
-2	1849	1870
-3	1871	1888
-4	1889	1899
-5	1900	1906
-6	1907	1912
-7	1913	1917
-8	1918	1922
-9	1923	1926
-10	1927	1930
-11	1931	1934
-12	1935	1938
-13	1939	1941
-14	1942	1944
-15	1945	1947
-16	1948	1950
-17	1951	1953
-18	1954	1956
-19	1957	1959
-20	1960	1962
-21	1963	1965
-22	1966	1967
-23	1968	1969
-24	1970	1971
-25	1972	1973
-26	1974	1975
-27	1976	1977
-28	1978	1979
-29	1980	1981
-30	1982	1983
-31	1984	1985
-32	1986	1987
-33	1988	1989
-34	1990	1991
-35	1992	1993
-36	1994	1994
-37	1995	1996
-38	1997	1997
-39	1998	1998
-40	1999	1999
-41	2000	2000
-42	2001	2001
-43	2002	2002
-44	2003	2003
-45	2004	2004
-46	2005	2005
-47	2006	2006
-48	2007	2007
-49	2008	2008
-50	2009	2009
-51	2010	2010
-52	2011	2011
-53	2012	2012
-54	2013	2013
-55	2014	2014
-56	2015	2015
-57	2016	2016
-58	2017	2017
-59	2018	2018

data/yago11k_both/about.txt

@@ -1,15 +0,0 @@
-# triples: 78032 
-# entities: 10526 
-# relations: 177 
-# timesteps: 46 
-# test triples: 6909 
-# valid triples: 7198 
-# train triples: 63925 
-Measure method:  N/A  
-Target Size :  0  
-Grow Factor:  0  
-Shrink Factor:  0  
-Epsilon Factor: 5.0  
-Search method: N/A  
-filter_dupes: both
-nonames: False

data/yago11k_both/relations.dict

@@ -1,177 +0,0 @@
-0	<wasBornIn>[0-2]
-1	<wasBornIn>[2-5]
-2	<wasBornIn>[5-7]
-3	<wasBornIn>[7-10]
-4	<wasBornIn>[10-12]
-5	<wasBornIn>[12-15]
-6	<wasBornIn>[15-17]
-7	<wasBornIn>[17-20]
-8	<wasBornIn>[20-22]
-9	<wasBornIn>[22-25]
-10	<wasBornIn>[25-27]
-11	<wasBornIn>[27-30]
-12	<wasBornIn>[30-32]
-13	<wasBornIn>[32-35]
-14	<wasBornIn>[35-45]
-15	<wasBornIn>[52-52]
-16	<diedIn>[0-3]
-17	<diedIn>[3-5]
-18	<diedIn>[5-7]
-19	<diedIn>[7-10]
-20	<diedIn>[10-12]
-21	<diedIn>[12-14]
-22	<diedIn>[14-17]
-23	<diedIn>[17-19]
-24	<diedIn>[19-21]
-25	<diedIn>[21-23]
-26	<diedIn>[23-25]
-27	<diedIn>[25-27]
-28	<diedIn>[27-29]
-29	<diedIn>[29-32]
-30	<diedIn>[32-34]
-31	<diedIn>[34-36]
-32	<diedIn>[36-38]
-33	<diedIn>[38-40]
-34	<diedIn>[40-42]
-35	<diedIn>[42-44]
-36	<diedIn>[44-47]
-37	<diedIn>[47-49]
-38	<diedIn>[49-51]
-39	<diedIn>[51-53]
-40	<diedIn>[53-55]
-41	<diedIn>[55-57]
-42	<diedIn>[59-59]
-43	<worksAt>[0-3]
-44	<worksAt>[3-5]
-45	<worksAt>[5-7]
-46	<worksAt>[7-10]
-47	<worksAt>[10-12]
-48	<worksAt>[12-14]
-49	<worksAt>[14-17]
-50	<worksAt>[17-19]
-51	<worksAt>[19-21]
-52	<worksAt>[21-23]
-53	<worksAt>[23-25]
-54	<worksAt>[25-27]
-55	<worksAt>[27-29]
-56	<worksAt>[29-32]
-57	<worksAt>[32-34]
-58	<worksAt>[34-36]
-59	<worksAt>[36-40]
-60	<worksAt>[40-42]
-61	<worksAt>[42-47]
-62	<worksAt>[47-53]
-63	<worksAt>[59-59]
-64	<playsFor>[0-3]
-65	<playsFor>[3-5]
-66	<playsFor>[5-23]
-67	<playsFor>[23-25]
-68	<playsFor>[25-27]
-69	<playsFor>[27-29]
-70	<playsFor>[29-32]
-71	<playsFor>[32-34]
-72	<playsFor>[34-36]
-73	<playsFor>[36-38]
-74	<playsFor>[38-40]
-75	<playsFor>[40-42]
-76	<playsFor>[42-44]
-77	<playsFor>[44-47]
-78	<playsFor>[47-51]
-79	<playsFor>[59-59]
-80	<hasWonPrize>[1-4]
-81	<hasWonPrize>[4-6]
-82	<hasWonPrize>[6-8]
-83	<hasWonPrize>[8-11]
-84	<hasWonPrize>[11-15]
-85	<hasWonPrize>[15-18]
-86	<hasWonPrize>[18-22]
-87	<hasWonPrize>[22-26]
-88	<hasWonPrize>[26-30]
-89	<hasWonPrize>[30-33]
-90	<hasWonPrize>[33-37]
-91	<hasWonPrize>[37-47]
-92	<hasWonPrize>[47-53]
-93	<hasWonPrize>[59-59]
-94	<isMarriedTo>[0-3]
-95	<isMarriedTo>[3-5]
-96	<isMarriedTo>[5-7]
-97	<isMarriedTo>[7-10]
-98	<isMarriedTo>[10-12]
-99	<isMarriedTo>[12-14]
-100	<isMarriedTo>[14-17]
-101	<isMarriedTo>[17-19]
-102	<isMarriedTo>[19-21]
-103	<isMarriedTo>[21-23]
-104	<isMarriedTo>[23-25]
-105	<isMarriedTo>[25-27]
-106	<isMarriedTo>[27-29]
-107	<isMarriedTo>[29-32]
-108	<isMarriedTo>[32-34]
-109	<isMarriedTo>[34-38]
-110	<isMarriedTo>[38-42]
-111	<isMarriedTo>[42-47]
-112	<isMarriedTo>[47-51]
-113	<isMarriedTo>[51-55]
-114	<isMarriedTo>[59-59]
-115	<owns>[0-10]
-116	<owns>[10-17]
-117	<owns>[17-19]
-118	<owns>[19-23]
-119	<owns>[23-36]
-120	<owns>[36-38]
-121	<owns>[59-59]
-122	<graduatedFrom>[0-3]
-123	<graduatedFrom>[3-5]
-124	<graduatedFrom>[5-7]
-125	<graduatedFrom>[7-10]
-126	<graduatedFrom>[10-14]
-127	<graduatedFrom>[14-17]
-128	<graduatedFrom>[17-19]
-129	<graduatedFrom>[19-21]
-130	<graduatedFrom>[21-23]
-131	<graduatedFrom>[23-27]
-132	<graduatedFrom>[27-32]
-133	<graduatedFrom>[32-34]
-134	<graduatedFrom>[34-38]
-135	<graduatedFrom>[38-42]
-136	<graduatedFrom>[59-59]
-137	<isAffiliatedTo>[1-4]
-138	<isAffiliatedTo>[4-6]
-139	<isAffiliatedTo>[6-8]
-140	<isAffiliatedTo>[8-11]
-141	<isAffiliatedTo>[11-13]
-142	<isAffiliatedTo>[13-15]
-143	<isAffiliatedTo>[15-18]
-144	<isAffiliatedTo>[18-20]
-145	<isAffiliatedTo>[20-22]
-146	<isAffiliatedTo>[22-24]
-147	<isAffiliatedTo>[24-26]
-148	<isAffiliatedTo>[26-28]
-149	<isAffiliatedTo>[28-30]
-150	<isAffiliatedTo>[30-33]
-151	<isAffiliatedTo>[33-35]
-152	<isAffiliatedTo>[35-37]
-153	<isAffiliatedTo>[37-40]
-154	<isAffiliatedTo>[40-42]
-155	<isAffiliatedTo>[42-44]
-156	<isAffiliatedTo>[44-47]
-157	<isAffiliatedTo>[47-49]
-158	<isAffiliatedTo>[49-51]
-159	<isAffiliatedTo>[51-53]
-160	<isAffiliatedTo>[53-55]
-161	<isAffiliatedTo>[55-57]
-162	<isAffiliatedTo>[59-59]
-163	<created>[0-3]
-164	<created>[3-5]
-165	<created>[5-10]
-166	<created>[10-12]
-167	<created>[12-17]
-168	<created>[17-19]
-169	<created>[19-25]
-170	<created>[25-29]
-171	<created>[29-32]
-172	<created>[32-36]
-173	<created>[36-42]
-174	<created>[42-47]
-175	<created>[47-53]
-176	<created>[59-59]

data/yago11k_both/time_map.dict

@@ -1,60 +0,0 @@
-0	-431	1782
-1	1783	1848
-2	1849	1870
-3	1871	1888
-4	1889	1899
-5	1900	1906
-6	1907	1912
-7	1913	1917
-8	1918	1922
-9	1923	1926
-10	1927	1930
-11	1931	1934
-12	1935	1938
-13	1939	1941
-14	1942	1944
-15	1945	1947
-16	1948	1950
-17	1951	1953
-18	1954	1956
-19	1957	1959
-20	1960	1962
-21	1963	1965
-22	1966	1967
-23	1968	1969
-24	1970	1971
-25	1972	1973
-26	1974	1975
-27	1976	1977
-28	1978	1979
-29	1980	1981
-30	1982	1983
-31	1984	1985
-32	1986	1987
-33	1988	1989
-34	1990	1991
-35	1992	1993
-36	1994	1994
-37	1995	1996
-38	1997	1997
-39	1998	1998
-40	1999	1999
-41	2000	2000
-42	2001	2001
-43	2002	2002
-44	2003	2003
-45	2004	2004
-46	2005	2005
-47	2006	2006
-48	2007	2007
-49	2008	2008
-50	2009	2009
-51	2010	2010
-52	2011	2011
-53	2012	2012
-54	2013	2013
-55	2014	2014
-56	2015	2015
-57	2016	2016
-58	2017	2017
-59	2018	2018

data_loader.py

@@ -1,136 +0,0 @@
-import torch
-import numpy as np
-
-from torch.utils.data import Dataset
-
-
-class TrainDataset(Dataset):
-    """
-    Training Dataset class.
-    Parameters
-    ----------
-    triples:	The triples used for training the model
-    params:		Parameters for the experiments
-
-    Returns
-    -------
-    A training Dataset class instance used by DataLoader
-    """
-
-    def __init__(self, triples, params):
-        self.triples = triples
-        self.p = params
-        self.strategy = self.p.train_strategy
-        self.entities = np.arange(self.p.num_ent, dtype=np.int32)
-
-    def __len__(self):
-        return len(self.triples)
-
-    def __getitem__(self, idx):
-        ele = self.triples[idx]
-        triple, label, sub_samp = torch.LongTensor(ele['triple']), np.int32(
-            ele['label']), np.float32(ele['sub_samp'])
-        trp_label = self.get_label(label)
-
-        if self.p.lbl_smooth != 0.0:
-            trp_label = (1.0 - self.p.lbl_smooth) * \
-                trp_label + (1.0/self.p.num_ent)
-
-        if self.strategy == 'one_to_n':
-            return triple, trp_label, None, None
-
-        elif self.strategy == 'one_to_x':
-            sub_samp = torch.FloatTensor([sub_samp])
-            neg_ent = torch.LongTensor(self.get_neg_ent(triple, label))
-            return triple, trp_label, neg_ent, sub_samp
-        else:
-            raise NotImplementedError
-
-    @staticmethod
-    def collate_fn(data):
-        triple = torch.stack([_[0] for _ in data], dim=0)
-        trp_label = torch.stack([_[1] for _ in data], dim=0)
-
-        if not data[0][2] is None:							# one_to_x
-            neg_ent = torch.stack([_[2] for _ in data], dim=0)
-            sub_samp = torch.cat([_[3] for _ in data], dim=0)
-            return triple, trp_label, neg_ent, sub_samp
-        else:
-            return triple, trp_label
-
-    def get_neg_ent(self, triple, label):
-        def get(triple, label):
-            if self.strategy == 'one_to_x':
-                pos_obj = triple[2]
-                mask = np.ones([self.p.num_ent], dtype=np.bool)
-                mask[label] = 0
-                neg_ent = np.int32(np.random.choice(
-                    self.entities[mask], self.p.neg_num, replace=False)).reshape([-1])
-                neg_ent = np.concatenate((pos_obj.reshape([-1]), neg_ent))
-            else:
-                pos_obj = label
-                mask = np.ones([self.p.num_ent], dtype=np.bool)
-                mask[label] = 0
-                neg_ent = np.int32(np.random.choice(
-                    self.entities[mask], self.p.neg_num - len(label), replace=False)).reshape([-1])
-                neg_ent = np.concatenate((pos_obj.reshape([-1]), neg_ent))
-
-                if len(neg_ent) > self.p.neg_num:
-                    import pdb
-                    pdb.set_trace()
-
-            return neg_ent
-
-        neg_ent = get(triple, label)
-        return neg_ent
-
-    def get_label(self, label):
-        if self.strategy == 'one_to_n':
-            y = np.zeros([self.p.num_ent], dtype=np.float32)
-            for e2 in label:
-                y[e2] = 1.0
-        elif self.strategy == 'one_to_x':
-            y = [1] + [0] * self.p.neg_num
-        else:
-            raise NotImplementedError
-        return torch.FloatTensor(y)
-
-
-class TestDataset(Dataset):
-    """
-    Evaluation Dataset class.
-    Parameters
-    ----------
-    triples:	The triples used for evaluating the model
-    params:		Parameters for the experiments
-
-    Returns
-    -------
-    An evaluation Dataset class instance used by DataLoader for model evaluation
-    """
-
-    def __init__(self, triples, params):
-        self.triples = triples
-        self.p = params
-
-    def __len__(self):
-        return len(self.triples)
-
-    def __getitem__(self, idx):
-        ele = self.triples[idx]
-        triple, label = torch.LongTensor(ele['triple']), np.int32(ele['label'])
-        label = self.get_label(label)
-
-        return triple, label
-
-    @staticmethod
-    def collate_fn(data):
-        triple = torch.stack([_[0] for _ in data], dim=0)
-        label = torch.stack([_[1] for _ in data], dim=0)
-        return triple, label
-
-    def get_label(self, label):
-        y = np.zeros([self.p.num_ent], dtype=np.float32)
-        for e2 in label:
-            y[e2] = 1.0
-        return torch.FloatTensor(y)

dataset/FB15k-237/.ipynb_checkpoints/get_neighbor-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 2
+}