智慧型OPAC主題檢索系統設計與實作

Abstract

鑑於線上公用目錄主題檢索失敗率非但居高不下，且一般檢索者對於如何進行主題檢索也普遍缺乏基本概念，本論文以實作系統解決問題，同時驗證查全率與查準率的關係是否能夠同時取得令人滿意的結果。 本系統的核心概念與作法是以自由詞彙(Free Text)為檢索詞，選擇學科範圍並設定不同的檢索準確等級後，檢索書目檔內與主題相關的欄位。系統除直接顯示檢索結果外，並提供檢索者自行選擇檢索結果書目中所含合適的標題與類號執行檢索，或由系統為檢索者自動選擇合適的標題與類號執行檢索，換言之，本系統可由自由詞彙找出適當的標題與類號作為檢索條件。經實驗結果驗證，設計良好的系統可降低主題檢索失敗率，且查全率與查準率亦可同時獲致良好的結果。

For a long time, the failure rate of OPAC (Online Public Access Catalog) subject search is tremendously high, and many searchers have no idea about how to use the subject search, In view of this, this thesis proposes an intelligent system for OPAC subject search to solve this problem. Through this system, we also want to verify if both of the recall ratio and precision ratio can be satisfactory. The core concept and method of the proposed system lies in discovering appropriate subject headings and classification numbers from free texts. A user chooses the subject category to be searched, sets an accuracy level, and submits free-text terms to search the subject-related fields in the underlying MARC file. The searchers can choose to display search result directly; moreover, they can choose their suitable headings and classification numbers in the search result, or the system will choose the suitable headings and classification numbers for the researcher automatically. According to our experiment, this system can effectively reduce the failure rate of OPAC subject search. Furthermore, both of the recall ratio and precision ratio can reach a satisfactory level at the same time.