arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.08145cs.CLcs.AIcs.LG
收藏

KI-BERT:注入知识语境以更好地理解语言和领域

KI-BERT: Infusing Knowledge Context for Better Language and Domain Understanding

Embibe;University of South Carolina(南卡罗来纳大学)

Keyur Faldu, Amit Sheth, Prashant Kikani, Hemang Akbari

摘要:

由最先进的基于转换器的语言模型(TLM)(如BERT、GPT、T5等)学习的上下文实体表示,利用注意力机制从训练数据语料库中学习数据上下文。然而,这些模型不使用知识背景。知识上下文可以理解为知识图谱中实体及其与相邻实体之间关系的语义。我们提出了一种新的和有效的技术,注入知识上下文从多个知识图谱的概念和模糊的实体到TLM在微调。该算法将知识图谱嵌入到齐次向量空间中,为实体引入新的标记类型,对齐实体位置ID,并引入选择性注意力机制。我们以BERT为基准模型,通过注入ConceptNet和WordNet中的知识上下文实现了“知识注入的BERT”,在GLUE基准测试的8个不同子任务上,该模型的性能明显优于BERT和其他最近的知识感知BERT变体,如ERNIE,SenseBERT和BERT_CS。KI-BERT-base模型甚至在特定领域的任务(如SciTail和QQP、QNLI和MNLI的学术子集)上的表现明显优于BERT-large。

英文摘要:

Contextualized entity representations learned by state-of-the-art transformer-based language models (TLMs) like BERT, GPT, T5, etc., leverage the attention mechanism to learn the data context from training data corpus. However, these models do not use the knowledge context. Knowledge context can be understood as semantics about entities and their relationship with neighboring entities in knowledge graphs. We propose a novel and effective technique to infuse knowledge context from multiple knowledge graphs for conceptual and ambiguous entities into TLMs during fine-tuning. It projects knowledge graph embeddings in the homogeneous vector-space, introduces new token-types for entities, aligns entity position ids, and a selective attention mechanism. We take BERT as a baseline model and implement the "Knowledge-Infused BERT" by infusing knowledge context from ConceptNet and WordNet, which significantly outperforms BERT and other recent knowledge-aware BERT variants like ERNIE, SenseBERT, and BERT_CS over eight different subtasks of GLUE benchmark. The KI-BERT-base model even significantly outperforms BERT-large for domain-specific tasks like SciTail and academic subsets of QQP, QNLI, and MNLI.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑