藉頭字語與引用文分析追蹤技術發展

Abstract

學術發展中的每樣技術，皆有不斷進化或衍生的可能性。當研究者因實驗需求而去探討某項技術時，往往必須耗費時間心力在閱覽大量學術文獻，以了解該項技術的發展境況。本論文的研究目的在於探討如何讓電腦利用頭字語和引用文(citation)提供的資訊，來對文件進行篩選並提取出有用的訊息，以幫助研究者有效率地掌握需求技術的後續發展脈絡。我們利用學術搜尋引擎Google Scholar的“被引用次數”搜尋功能，來找出所有與需求技術文件相關的學術文件，建立了三組測試資料組，並分別以人工閱讀方式找出符合我們實驗目的的文件集合，作為衡量實驗結果用的比對資料。在這份論文中，文件分析過程主要可分為三個主要部份：1.從文件中擷取並篩選頭字語、2.利用頭字語和需求技術關鍵字來對蒐集的文件進行篩選、3.對篩選後的文件進行引用文及共引文(co-citation)分析。我們將在論文後段從相關技術資訊取得和有用文件取得兩方面，來探討論文的實驗結果。

Each technology developed in academic research has the possibility for further improvement or extension in the future. When a researcher investigates a certain technology due to the requirement of his/her own work, it usually takes a lot of time and effort for that researcher in surveying the literature to know the current status of that technology. The purpose of this thesis is to investigate how the information provided by acronyms and citations in the documents can be utilized for the automatic screening of documents and the extraction of helpful information. The goal is to help researchers more efficiently grasp the current development of the technology they need. We use the “Cited by” function of the academic search engine Google Scholar to find all the scholarly literature related to some given technical documents and established three sets of test data. To be able to evaluate our method, we manually read through all the citing documents to find the ones that are consistent with our purpose. In this paper, document analysis consists of three main parts: (1) The extraction and selection of acronyms from the documents; (2) the use the acronyms and keywords of related to the target technology to automatically screen the collection of citing documents; (3) citation analysis and co-citation analysis for the selected citing documents. We will discuss our experimental results on two aspects: to retrieve useful information about the target technology, and the selection of citing documents that fit our purpose.