圖書館的書目編目工作人員需要花費大量的時間對新進書籍進行編目,根據過去經驗與規則,給予每本書適當的分類號與主題詞等。本研究提出一套自動化的系統,能夠自動推薦適合的分類號與主題詞,作為編目人員的參考,加速編目工作流程。 方法上利用文件檢索的概念,將既有的編目資料建成索引,用搜尋的方式提供初步的相關詞條,最後經過重新計算與排序給出最終的推薦結果。系統流程上主要分成:資料前處理、建立索引、搜尋排序、例外規則分類四大部份,針對圖書館編目資料的特性設計適合的處理方式。 本研究的目的在於讓圖書館方編目人員在新書編目的過程中,能夠以系統推薦的分類號與主題詞加速編目工作流程。此外從研究過程中,進一步發現可研究的議題或更有價值的資訊。
Librarians spend a lot of time and efforts on new book bibliography. They choose moderate classification and subject headings for new books according to the rule from the manual and the experience from their career. This research offers an approach to build an automatic recommendation system to suggest classification codes and subject headings of new books. By the concept of information retrieval, we built the index of the bibliography data to search for the candidate classification codes and subject headings. After that, the system sort the candidate list to find the recommended results. The whole system is divided into 4 parts: data pre-process, data indexing, candidate retrieval & ranking process, and rule classification. The goal of this research is to help librarians’ daily work for new book bibliography and increase the efficiency of their works. Besides, the research also finds out some interesting issues for recommendation systems applied for the real library data set.