EAPSI: Investigating a Novel Approach for Biological Named Entity Recognition in Text Mining

目前，使用计算资源从生物文献中提取信息面临着许多挑战，由于生物文本的领域特定术语。为了阅读和获取信息，从大量的生物学文献，需要大量的资金，时间和人力。研究文献和数据的数量一直在增加，需要一个可扩展的解决方案来从生物文本中收集信息。本项目旨在开发一种从生物学文献中提取特定注释的新技术。在本项目中，主要研究者将前往位于台湾台北的中央研究院与徐文莲博士合作，徐文莲博士在自然语言处理和生物文献文本挖掘领域的专业知识对于实施该方法是必要的。DNA甲基化被认为是诊断和治疗癌症的潜在生物标志物。最近的研究已经确定了异常基因甲基化和癌症发展之间的关系。在以前的生物学文献挖掘方法中，人工策展和基于机器学习的方法的组合已被用于提取基因甲基化与癌症的关系。该项目将利用一种新的基于统计模式的方法，重点是识别疾病术语和物理相互作用的深层和重要的语言模式。这种方法具有独特的灵活匹配机制，可以捕获文本中现有的疾病术语或蛋白质名称，而不管它们之前是否存在于训练数据中。该项目解决了在短时间内阅读和分析大量生物学文献的资源限制问题。该项目是由美国国家科学基金会和台湾科技部共同资助的东亚和太平洋暑期研究所项目下的一个奖项，用于支持美国研究生的暑期研究。

Currently, using computational resources to extract information from biological literature faces many challenges due to the domain specific terminology of biological texts. In order to read and obtain information from the vast amount of available biological literature, a large amount of funds, time and manpower are required. The amount of research literature and data is always increasing, and a scalable solution to gleaning information from biological text is necessary. This project aims to develop a new technique for extracting specific annotations from biological literature. For this project, the PI will travel to Academia Sinica in Taipei, Taiwan to work with Dr. WenLian Hsu, whose expertise in the areas of natural language processing and text mining of biological literature is necessary for the implementation of this approach.DNA methylation is regarded as a potential biomarker in the diagnosis and treatment of cancer. Recent studies have identified relationships between aberrant gene methylation and cancer development. In previous approaches of mining biological literature, a combination of manual curation and machine learning-based approaches have been used to extract gene methylation cancer relations. This project will utilize a novel statistical pattern based approach which focuses on identifying deep and important linguistic patterns described for disease terms and physical interactions. This approach has a unique flexible matching mechanism that captures existing disease terms or protein names within the text, regardless of their previous presence in the training data. This project provides a solution to the resource limitation of reading and analyzing large amounts of biological literature in a short amount of time.This award under the East Asia and Pacific Summer Institutes program supports summer research by a U.S. graduate student and is jointly funded by NSF and the Ministry of Science and Technology of Taiwan.