Advanced End-to-End Relation Extraction with Deep Neural Networks

使用深度神经网络进行高级端到端关系提取

基本信息

批准号：
10615695
负责人：
Venkata Naga Ramakanth Kavuluru
金额：
$ 33.27万
依托单位：
UNIVERSITY OF KENTUCKY
依托单位国家：
美国
项目类别：
财政年份：
2020
资助国家：
美国
起止时间：
2020-07-01 至 2025-03-31
项目状态：
未结题

来源：
https://reporter.nih.gov/project-details/10615695
关键词：
Adverse event Architecture Area Benchmarking Bioinformatics Biology Biomedical Research Black race Classification Clinical Code Collaborations Combination Drug Therapy Communication Communities Complex Computer software Data Data Set Dependence Disease Distant Drug Interactions Encapsulated Etiology Evaluation Fostering Funding Future Generations Genes Growth Hand Heart Information Retrieval Information Sciences Intramural Research Joints Knowledge Discovery Label Language Lead Link Literature Manuals Maps Measures Methodology Methods Modeling Molecular Names Natural Language Processing Patients Peer Review Performance Periodicals Pharmaceutical Preparations Physicians Positioning Attribute Process Reporting Research Research Personnel Resources Review Literature Scientist Semantics Software Tools Source Standardization Structure Supervision System Terminology Testing Text Training Translational Research Trees United States National Library of Medicine biomedical data science clinical care deep neural network improved insight interest knowledge base knowledgebase machine learning method natural language neural neural network neural network architecture new therapeutic target novel off-label use protein protein interaction side effect social media supervised learning syntax transfer learning

项目摘要

ABSTRACT Relations linking various biomedical entities constitute a crucial resource that enables biomedical data science applications and knowledge discovery. Relational information spans the translational science spectrum going from biology (e.g., protein–protein interactions) to translational bioinformatics (e.g., gene–disease associations), and eventually to clinical care (e.g., drug–drug interactions). Scientists report newly discovered relations in nat- ural language through peer-reviewed literature and physicians may communicate them in clinical notes. More recently, patients are also reporting side-effects and adverse events on social media. With exponential growth in textual data, advances in biomedical natural language processing (BioNLP) methods are gaining prominence for biomedical relation extraction (BRE) from text. Most current efforts in BRE follow a pipeline approach containing named entity recognition (NER), entity normalization (EN), and relation classiﬁcation (RC) as subtasks. They typically suffer from error snowballing — errors in a component of the pipeline leading to more downstream errors — resulting in lower performance of the overall BRE system. This situation has lead to evaluation of different BRE substaks conducted in isolation. In this proposal we make a strong case for strictly end-to-end evaluations where relations are to be produced from raw text. We propose novel deep neural network architectures that model BRE in an end-to-end fashion and directly identify relations and corresponding entity spans in a single pass. We also extend our architectures to n-ary and cross-sentence settings where more than two entities may need to be linked even as the relation is expressed across multiple sentences. We also propose to create two new gold standard BRE datasets, one for drug–disease treatment relations and another ﬁrst of a kind dataset for combination drug therapies. Our main hypothesis is that our end-to-end extraction models will yield supe- rior performance when compared with traditional pipelines. We test this through (1). intrinsic evaluations based on standard performance measures with several gold standard datasets and (2). extrinsic application oriented assessments of relations extracted with use-cases in information retrieval, question answering, and knowledge base completion. All software and data developed as part of this project will be made available for public use and we hope this will foster rigorous end-to-end benchmarking of BRE systems.

抽象的连接各种生物医学实体的关系构成了实现生物医学数据科学的重要资源应用和知识发现跨越了转化科学领域。从生物学（例如蛋白质-蛋白质相互作用）到转化生物信息学（例如基因-疾病关联），并最终进入临床护理（例如，药物与药物的相互作用）。通过同行评审的文献来了解口头语言，医生可以在临床笔记中传达它们。最近，患者也在社交媒体上报告副作用和不良事件，数量呈指数级增长。文本数据，生物医学自然语言处理（BioNLP）方法的进步正在引起人们的关注从文本中提取生物医学关系（BRE）目前大多数 BRE 工作都遵循包含以下内容的管道方法。命名实体识别（NER）、实体标准化（EN）和关系分类（RC）作为子任务。通常会遭受错误滚雪球般的影响——管道组件中的错误会导致更多下游错误 ——导致整个BRE系统的性能降低，这种情况导致了不同的评估。在此提案中，我们强烈建议进行严格的端到端评估。我们提出了新颖的深度神经网络架构，其中关系是从原始文本中产生的。以端到端的方式对 BRE 进行建模，并直接识别单个关系和相应的实体跨度我们还将我们的架构扩展到 n 元和跨句子设置，其中可能有两个以上的实体。即使关系是跨多个句子表达的，我们也建议创建两个。新的黄金标准 BRE 数据集，一个用于药物与疾病治疗关系的数据集，另一个同类数据集我们的主要假设是我们的端到端提取模型将产生超强的效果。与传统管道相比，我们通过基于（1）的内在评估来测试这一点。基于多个黄金标准数据集的标准性能测量和（2）。对信息检索、问答和知识中的用例提取的关系进行评估作为该项目一部分开发的所有软件和数据将可供公众使用和使用。我们希望这将促进 BRE 系统严格的端到端基准测试。

项目成果

期刊论文数量（7）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

Improved biomedical word embeddings in the transformer era.

改进了 Transformer 时代的生物医学词嵌入。

DOI：
发表时间：
2021-08
期刊：
影响因子：
4.5
作者：
Noh, Jiho;Kavuluru, Ramakanth
通讯作者：
Kavuluru, Ramakanth

Joint Learning for Biomedical NER and Entity Normalization: Encoding Schemes, Counterfactual Examples, and Zero-Shot Evaluation.

生物医学 NER 和实体标准化的联合学习：编码方案、反事实示例和零样本评估。

DOI：
发表时间：
2021-08
期刊：
影响因子：
0
作者：
Noh, Jiho;Kavuluru, Ramakanth
通讯作者：
Kavuluru, Ramakanth

End-to-End Models for Chemical-Protein Interaction Extraction: Better Tokenization and Span-Based Pipeline Strategies.

化学-蛋白质相互作用提取的端到端模型：更好的标记化和基于跨度的管道策略。

DOI：
发表时间：
2023-06
期刊：
影响因子：
0
作者：
Ai, Xuguang;Kavuluru, Ramakanth
通讯作者：
Kavuluru, Ramakanth

Acquisition of a Lexicon for Family History Information: Bidirectional Encoder Representations From Transformers-Assisted Sublanguage Analysis.

获取家族史信息词典：来自变压器辅助子语言分析的双向编码器表示。

DOI：
发表时间：
2023-06-27
期刊：
影响因子：
3.2
作者：
Wang, Liwei;He, Huan;Wen, Andrew;Moon, Sungrim;Fu, Sunyang;Peterson, Kevin J;Ai, Xuguang;Liu, Sijia;Kavuluru, Ramakanth;Liu, Hongfang
通讯作者：
Liu, Hongfang

COVID-19 Event Extraction from Twitter via Extractive Question Answering with Continuous Prompts.

通过带有连续提示的提取问答从 Twitter 中提取 COVID-19 事件。