基于优化覆盖算法的异构多模态信息检索方法  

Heterogeneous Multimodal Information Retrieval Method Based on Optimal Coverage Algorithm

作　　者：衣天龙 邵琦 李秋元 YI Tian-long;SHAO Qi;LI Qiu-yuan(Wuhan University,Hubei Wuhan 430000,China)

机构地区：[1]武汉大学,湖北武汉430000

出　　处：《计算机仿真》2024年第10期493-496,506,共5页Computer Simulation

基　　金：国家自然科学基金重点国际(地区)合作项目(71420107026);科技部国家重点研发计划项目(2018YFC0806904-03)。

摘　　要：针对信息检索过程易受信息差异度、边缘数据等问题的干扰,导致检索结果的查全率和效率较低等问题,提出基于优化覆盖算法的异构多模态信息检索方法。根据邻量化距离计算异构多模态信息之间的差异度,根据不同差异度,采用张量分解聚类算法对信息实行聚类处理,采用粒子群优化覆盖算法在迭代过程中搜索出最佳的覆盖粒子,得到搜索效果最佳的覆盖值,完成异构多模态信息的检索。实验结果表明,所提方法的检索时间短、查全率高、NDCG数值高,有效提升了信息检索质量。A heterogeneous multimodal information retrieval method based on optimized coverage algorithm is proposed to address the interference of information differences,edge data,and other issues in the information retrieval process,resulting in low recall and efficiency of retrieval results.Firstly,the difference between heterogeneous multimodal information was calculated by the nearest neighbor quantization distance.According to different difference,the tensor decomposition algorithm was used for clustering the information.Then,the particle swarm optimization coverage algorithm was used to search for the best coverage particles during the iteration,and thus to obtain the coverage value with the best search effect.Finally,the retrieval of heterogeneous multi-modal information was completed.The experimental results prove that the retrieval time of the proposed method is shorter,and the recall rate is high.In addition,the NDCG value is high,which effectively improves the quality of information retrieval.

关 键 词：信息差异度 张量分解 基因网络 粒子群优化覆盖 适应度函数 信息投影 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]