生物醫學文獻知識視覺化呈現

Abstract

近年來由於科技資訊發達的影響下，生物醫學界每天都有大量資訊的電子文件刊登於各種不同的期刊上，然而面對大量且複雜的生物醫學文獻資料，要從資料中找出特定且對於研究人員有意義的資訊可說是非常重要但卻不容易，若能減少研究人員對於這些文獻資料處理所花費的時間，也就是間接的節省人力、物力以及時間，故如何從文件中擷取出相關知識勢必將成為問題的關鍵。 那又該從啥麼樣的資料著手？又該怎樣的處理資料才有效率？處理完畢後又該如何加以呈現？對於生物醫學文獻處理與知識擷取，傳統的文件探勘技術已經不敷使用，本研究將導入「基因本體論 (Gene Ontology) 」的概念，對於在生物醫學文獻中關於基因功能性描述 (Functional Profile) 的知識加以擷取、處理與呈現。本研究主要分為幾個階段：首先是針對原有的 Gene Ontology 資訊加以處理，再來則是從生物醫學文獻中擷取出相關資訊進而加以呈現，並且設計一個圖形化使用者介面系統 - OntoMarker ，讓研究人員透過簡單的操作方式，即可萃取出存在於單篇文獻或某一主題中的重要資訊。最後並以在 PubMed / MEDLINE 中有關 CARD15/NOD2 基因的醫學文獻來驗証本系統，驗証結果確實可行，對於研究人員有所幫助。

Under the influence of developed information technology in these years, biomedical circles have numerous e-documents published in various periodicals everyday; however, it is very importance but difficult to find out specific and meaningful information to researchers from numerous and complicated biomedical literatures. Therefore, if the processing time of these literatures by the researchers could be reduced, it is to save manpower, material and time indirectly; thus how to extract relevant knowledge from documents will certainly become the key of the problem. But what document shall begin with? How to process data effectively? How to present after process? For the process of biomedical literatures and extract of knowledge, the traditional document mining technology is no longer sufficient. Therefore, this research will introduce the concept of 'Gene Ontology' to extract, process, and present the knowledge of describing gene function in biomedical literatures. This research is divided to several stages: first is to process the original information of Gene Ontology, and then is to extract and present relevant information from biomedical literatures, and design a graphic user interface system – OntoMarker to enable researchers to extract the important information that existed in single literature or certain topic through easy operation. Finally is to verify the system by the medical literatures relevant to CARD15/NOD2 gene in PubMed/MEDLINE. The verified results are true and feasible, and helpful to researchers.