基于深度学习框架下的多模态情报智能挖掘研究  

Research on Multimodal Intelligence Mining Based on Deep Learning Framework

作　　者：王静茹 宋绍成 徐慧 WANG Jing-ru;SONG Shao-cheng;XU-Hui(School of Computer Science and Technology,Beihua University,Jilin 132013,CA/ntz)

机构地区：[1]北华大学计算机科学技术学院

出　　处：《情报科学》2019年第12期159-165,共7页Information Science

基　　金：吉林省教育厅重点教研项目“大学计算机泛在学习模式(4A)研究与实践”;国家社会科学基金项目“突发事件中多模态危机情报智能采集加工生成应对策略研究”(19BTQ057);吉林省公共安全研究基地成果

摘　　要：【目的/意义】本文基于深度学习的理论框架,以突发事件中多模态危机情报智能采集加工生成应对策略的过程作为研究对象,构建出第三视角视频采集模型,并按照"数据→信息→知识→情报"的递进顺序建立危机情报多层次多阶段智能化模型。【方法/过程】提供不同层次多模态情报挖掘智能方法,使其达到对突发事件中多模态危机数据智能采集后,通过智能挖掘分析把危机信息加工成危机知识并利用其智能生成突发事件分阶段应对策略,从而最终构建多模态危机情报智能管理体系。【结果/结论】本文的研究架构和情报智能挖掘,可为突发事件应急决策提供科学、客观的实践参考和理论指导。【Purpose/significance】In this article, the collection model from the third perspective is proposed, based on the theoretical framework of the in-depth learning, and focusing on coping strategy of intelligent collection and processing of multimodal crisis intelligence in the emergency. The multi-stage intelligent model was set up according to the progressive sequence of "data-information-knowledge-intelligence".【Method/process】By using the Literature measurement analysis method, in the CNKI database, searching through "the national natural science fund", "output papers", "effect" as keywords, download the papers about the national natural science fund to support the effect of research literature, according to the national natural science foundation funding for outputpapers, quality classification, carding the research status quo, respectively, analysis of existing problems and the reasons and countermeasures.【Result/conclusion】Current national natural science fund of fund effect research of output papers in the existence of unreasonable evaluation results, future research can control interference factors in the confusion of function under the premise of fund, compared with single and multiple support of NSFC major scientific fund at home and abroad in support role, further reasonable evaluating NSFC funding for output papers.

关 键 词：深度学习 视频采集 智能计算 知识挖掘 应急管理 

分 类 号：G250.2[文化科学—图书馆学]