基于層次聚類的圖書元數(shù)據(jù)語義聚合研究

打開文本圖片集

[關(guān)鍵詞]圖書元數(shù)據(jù)層次聚類BERT模型語義相似度語義距離

[中圖分類號]G250.7

[文獻(xiàn)標(biāo)志碼］A[DOI]10.19764/j.cnki.tsgjs.20240420

[本文引用格式]，.基于層次聚類的圖書元數(shù)據(jù)語義聚合研究[J].圖書館建設(shè)，2025（1）：82-93.

[Abstract]Achievingdeepfusionofheterogeneousbookresourcesfrommultiplesourcesiscrucialforexpandingthebreadthand comprehensivenessoflbrarysericesandpromotingtedevelopmentofinteligntlibarysystems.Amongthesechalengs，tesmantic agregationofdiverseanddiferentlynamedbookmetadataplaysapivotalroleinfaclitingtheepintegrationofbookinfmationfrom varioussurstdadalntfciabts Chineseisusedasawordembedingmodel.Fromtheperspectivesofmetadatafieldsthemselvesandtributevalues，semantisimilarity anddistancebtweenmetadataareanalyzd.Itisfundtatbasedndistanematrixierarchicalusteringcanbeachievedandang correspondencebetweenmetadatacanbeautomaticllyonstructed，therebyachievingsemanticagregationbetweenbookmetadataih similar names or atributes.Experimental results demonstrate an impressive mapping relationship precision rateof 93.33% ，significantly reducing thehumanefortrequiredduringthemetadataaggegationandfusionprocess.Furthermore，theproposedsemanticaggegationappoach forbookmetadataehibitsextensivealicabityalowingitsiterativreuseinotheriformationagregationsenaroswilmintaining compatibility and generality.

[Keywords]Book metadata; Hierarchical clustering; BERTmodel; Semantic similarity; Semantic distance

0引言

圖書作為知識的實際載體與傳播介質(zhì)，廣泛分布在不同的檢索發(fā)現(xiàn)平臺，雖然通常在各個平臺之間預(yù)留了訪問獲取鏈接，但由于它們在不同平臺的元數(shù)據(jù)格式名稱并不統(tǒng)一，使其在內(nèi)容層面的關(guān)聯(lián)整合層次較淺，呈現(xiàn)出了“聯(lián)而不合”的局面。（剩余13769字）