Knowledge graph-driven mining of emotional attribution in a centennial overseas Chinese poetry

Abstract As a significant medium of literary art, overseas Chinese poetry not only conveys the personal emotions of poets but also reflects the social ethos and cultural landscape of specific eras, serving as a century-spanning legacy of nostalgia and memory for overseas Chinese. To address the problems of fragmented emotional semantics and ambiguous emotional attribution correlation in centennial overseas Chinese poetry, this study proposes a knowledge graph (KG) construction framework centered on the emotional context attention mechanism. Through the emotional semantic fusion encoder and poetic entity decoder, the triple of emotion–imagery–author is accurately extracted, and the KG of emotional attribution of overseas Chinese poetry is constructed. The system can realize intelligent retrieval, emotional correlation analysis and trend mining, providing a replicable digital humanities method for the research of overseas Chinese hometown literature and homesickness.