当前位置：首页 > 过刊浏览->2023年45卷5期
推荐算法的船舶电子海图数据相似性检索方法
Similarity retrieval method of ship electronic chart data based on recommended algorithm
- DOI:
- 作者:
- 金仕奇
JIN Shi-qi
- 作者单位:
- 江西工程学院 大数据与计算机学院，江西 新余 338000
Jiangxi University of Engineering, Big Data and Computer College, Xinyu 338000, China
- 关键词:
- 推荐算法;船舶电子海图;数据相似性;检索方法;协同过滤;动态加权
recommended algorithm; ship electronic chart; data similarity; search method; collaborative filtering; dynamic weighting
- 摘要:
- 以提升船舶导航路线制定质量为目的，提出推荐算法的船舶电子海图数据相似性检索方法，提高电子海图数据相似性检索质量。通过加权核范数算法填充船舶电子海图数据稀疏评分矩阵，通过协同过滤推荐算法计算检索目标与电子海图数据间的评分相似性以及类别相似性；根据评分相似性与类别相似性，得到基于评分相似性与类别相似性的电子海图数据推荐预测评分，以动态加权方式得到电子海图数据相似性检索结果。实验证明：该方法可有效完成船舶电子海图数据相似性检索；不同电子海图数据评分矩阵稀疏度时，该方法电子海图数据相似性检索的归一化折损累积增益均较高，即检索精度较高。
In order to improve the quality of ship navigation route, a similarity retrieval method of ship electronic chart data based on recommended algorithm is proposed to improve the quality of similarity retrieval of electronic chart data. The sparse score matrix of ship electronic chart data is filled by weighted kernel norm algorithm, and the score similarity and category similarity between the retrieval target and electronic chart data are calculated by collaborative filtering recommendation algorithm. According to the score similarity and category similarity, the electronic chart data recommendation prediction score based on the score similarity and category similarity is obtained, and the electronic chart data similarity retrieval results are obtained by dynamic weighting. Experiments show that this method can effectively complete the similarity retrieval of ship electronic chart data. When the sparsity of the scoring matrix of the electronic chart data is different, the normalized cumulative loss gain of the similarity retrieval of the electronic chart data is higher, that is, the retrieval accuracy is higher.
2023,45(5): 148-151 收稿日期：2022-10-16
DOI：10.3404/j.issn.1672-7649.2023.05.028
分类号：TP391
作者简介：金仕奇(1982-),男,讲师,主要从事计算机科学与技术方向研究
