期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2018
卷号：7
期号：6
页码：6527-6533
DOI：10.15680/IJIRSET.2018.0706021
出版社：S&S Publications
摘要：Hashing methods are important and widely useful technique in recent years. Several methods it
introduced to find out the similarities between text, video-audio, and cross-media information. In existing, use a bag-ofwords
is a technique of representing text data and also used in natural language process-ing and information retrieval.
Because information with various forms may have a same meaning, semantic text similarities are not well elaborated in
these methods.
In this paper, proposed a new method called semantic cross media hashing (SCMH), which uses bag of words by
capturing the semantic textual similarity level. In proposed method, two commonly used cross media datasets (i.e.
mriflicker and label me).
关键词：Fisher vector; hashing method ;SIFT descriptor; word embedding;
