基于时间知识图谱嵌入的电力恐怖主义事件预测

打开文本图片集

关键词：恐怖主义；事件预测；时间知识图谱；门控循环单元；注意力机制DOI：10.15938/j. jhust. 2025.0.006中图分类号：P208 文献标志码：A 文章编号：1007-2683（2025）0-008-10

Electric Power Terrorism Event Prediction Based on Temporal Knowledge Graph Embedding

CHEN Hongshan¹， ZHOU Peng'， GAO Hongliang¹， YANG Zhengquan²， SHI Kan³， DING Bo4 （1. China Southern Power Grid Power Dispatching Control Center， Guangzhou 51O623，China;

2. Honghe Power Supply Bureau， Yunnan Power Grid Co.，Ltd.，Mengzi 6611Oo，China;

3. Dehong Power Supply Bureau，Yunnan Power Grid Co.，Ltd.，Mangshi 67840O，China;

4.School of Computer Science and Technology，Harbin Universityof Scienceand Technology，Harbin 150o80，China）

Abstract：Itiscrucial topredictpowerterorismevents forensuringpeople'squalityoflifeandsocialstability.Existing metods usetheglobal terrismdatabase（GTD）tobuildatwo-layerstatic knowledgegraphforpredictingpowerterrorism events.GTDonly containstedatadirectlyrelatedtoterorismeventsandlacksbackgroundinformation.Thetwo-layerstaticknowledgegraphcannot takeeventsasthecenter，makingitdificulttoextractthetemporalandspatialrelationshipsamongevents.Tosolvetheabove problems，this paper proposesan electric power terorism event prediction methodbasedon temporal knowledgegraph embedding， whichcanefectivelyminethedatainGTDand Wikipedia，andconstructatemporalknowledgegraph withathre-layerstructureof patern，eventanddata.Furthermore，thispaperusesanatentio-basedhistoricaleventsembeddingmoduletoencodeastevents， andadecodermoduleisdevelopedtopredictonmultipleaspectsofelectricpower teroristeventsOurmethodcanefectivelybtain thecorelationbetweeneventsndthetopologicalreltioshipbetweeneventsandatributes，enablingacuatepredictionsonmultiple aspects of terrorist events.

Keywords ： terrorism;event prediction ;temporal knowledge graph ;gated recurrent unit ;attention mechanism

0 引言

恐怖袭击自20世纪90年代以来，在全球范围内迅速蔓延，对全世界人民的生命财产安全产生严重威胁。（剩余15320字）