关键词: 数字人文, 书目提要, 多粒度, 文本语义融合, 古籍智能推荐

Abstract: [Purpose/Significance] In view of the current situation that it is difficult to realize multi-dimensional recommendations such as the responsible person, content, and subsequent evaluation in ancient book retrieval, this paper explores an intelligent recommendation method of ancient books based on multi-granularity semantic fusion of bibliographic summaries. This paper aims to enhance the efficiency and accuracy of researchers in retrieving similar ancient books. [Design/Methodology] Taking bibliographic summaries as bridges, this paper constructs semantic connections among different ancient books. We propose MGF method for ancient books recommendation based on multi-granularity fusion of semantic information from bibliographic summaries. By adaptively aligning and integrating word-granularity named entities with sentence-granularity structural functions, MGF improves the effect of chapter-granularity semantic characterization of bibliographic summaries, thereby constructing a multi-dimensional connection network among ancient books to enable precise recommendation. [Findings/Conclusion] This paper takes ancient agricultural books and their bibliographic summaries as the research objects, and confirms the effectiveness and domain applicability of the proposed method. The experimental results show that this method can effectively capture the deep semantic correlations among ancient books and achieve their multi-dimensional intelligent recommendation. [Originality/Value] It can assist users to quickly locate relevant ancient books in terms of responsible person, content, and evaluations, improving the search efficiency, and discover wider range of literature associations including those of lost ancient books.

Keywords: Digital humanities, Bibliographic summaries, Multi-granularity, Text semantic fusion, Intelligent recommendation of ancient books