PaECTER - 专利相似度模型

PaECTER（基于引文信息的专利嵌入Transformer模型）是一种专利相似度模型。该模型以Google的BERT for Patents作为基础模型，从专利文本中生成1024维的密集向量嵌入。这些向量封装了给定专利文本的语义本质，使其非常适用于各种与专利分析相关的下游任务。

论文：https://arxiv.org/pdf/2402.19411

应用场景

语义搜索
现有技术检索
聚类
专利地图绘制

使用方法（Sentence-Transformers）

当您安装了sentence-transformers后，使用此模型会变得非常简单：

pip install -U sentence-transformers

然后你可以像这样使用该模型：

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('HangZhou_Ascend/paecter')
embeddings = model.encode(sentences)
print(embeddings)

用法（openmind）

如果不使用sentence-transformers，您可以这样使用该模型：首先，将输入传递给transformer模型，然后必须在上下文单词嵌入之上应用正确的池化操作。

from openmind import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('HangZhou_Ascend/paecter')
model = AutoModel.from_pretrained('HangZhou_Ascend/paecter')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt', max_length=512)

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

评估结果

本模型的评估详情可参见我们的论文：PaECTER: Patent-level Representation Learning using Citation-informed Transformers

训练

模型训练参数如下：

数据加载器：

长度为 318750 的 torch.utils.data.dataloader.DataLoader，参数如下：

{'batch_size': 4, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

损失函数：

sentence_transformers.losses.CustomTripletLoss.CustomTripletLoss，参数如下：

  {'distance_metric': 'TripletDistanceMetric.EUCLIDEAN', 'triplet_margin': 1}

fit() 方法的参数：

{
    "epochs": 1,
    "evaluation_steps": 4000,
    "evaluator": "sentence_transformers.evaluation.TripletEvaluator.TripletEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 1e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 31875.0,
    "weight_decay": 0.01
}

完整模型架构

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False})
)

引用与作者

@misc{ghosh2024paecter,
      title={PaECTER: Patent-level Representation Learning using Citation-informed Transformers}, 
      author={Mainak Ghosh and Sebastian Erhardt and Michael E. Rose and Erik Buunk and Dietmar Harhoff},
      year={2024},
      eprint={2402.19411},
      archivePrefix={arXiv},
      primaryClass={cs.IR}
}