关键词: 地名沿革, 信息聚合, 事件检测, 中国历史地理信息系统, 计算人文

Abstract: Historical geographic data provide spatial-temporal context for events involving historical figures. The effective organization and utilization of such data can deepen the semantic associations of classical texts. Taking the Book of Han(Han Shu) as the research object, the study explores how historical geographic data can support the organization and analysis of historical documents. By utilizing the China Historical Geographic Information System(CHGIS), historical place names and administrative division evolution data are obtained to construct a cross-temporal place-name knowledge base. Based on the event information extraction paradigm, it compares the applicability of traditional pretrained language models and large language models in detecting event information in classical Chinese texts. The GijiBERT-BiLSTM-CRF model is employed to extract event sentences from the biographies in the Book of Han and to annotate the event types. Using classical Chinese part-of-speech tagging and rule-matching methods, the study further extracts place names from the event sentences. A neural network and the toponym knowledge base are then employed to link the place names to entities, aggregating event sentences that refer to the same place under different names. CHGIS is used to visualize the spatial distribution of different types of events. A total of 18,525 event sentences were extracted from the biographies in the Book of Han. Through toponym entity linking, 1,253 event sentences were spatially located. Using place names as retrieval points for event information increased the recall rate to 1.3 times the original level. Visualization based on digital maps also improved the readability of historical data. This study validates the potential of historical geographic data in organizing and analyzing event information in ancient texts, providing a practical methodological reference for spatio-temporal association mining in historical documents.

Key words: Place name evolution, Information aggregation, Event detection, China , Historical Geographic Information System(CHGIS), Digital humanities