基于word2vec的中文歌词关键词提取算法  

Chinese Lyrics' Keyword Extraction Algorithm Based on Word2vec

作　　者：蒙晓燕 殷雁君[1] MENG Xiao-yan;YIN Yan-jun(College of Computer, Inner Mongolia Normal University, Hohhot 010022,China)

机构地区：[1]内蒙古师范大学计算机学院,内蒙古呼和浩特010022

出　　处：《内蒙古师范大学学报（自然科学汉文版）》2018年第2期137-140,共4页Journal of Inner Mongolia Normal University(Natural Science Edition)

基　　金：内蒙古自治区高等学校科学研究项目基金资助(NJZY13047)

摘　　要：为了让用户根据歌词内容快速精准地检索音乐,提出一种基于word2vec的中文歌词关键词提取算法.算法运用word2vec将歌词表征为词向量,根据词向量计算词语之间的相似度,其次通过K-means聚类算法得到歌词关键词.同时与基于TFIDF、LDA模型的歌词关键词提取结果进行比较,发现从该算法得出的10个关键词中抽取与人工标注相同的2-5个时,准确率明显增加.In order to make users search music by lyrics rapidly and accurately,A Chinese lyrics’ keyword extraction algorithm based on word2 vec is proposed,which is a model for deep learning.Firstly,this algorithm characterizes lyrics as word vector by word2 vec.Secondly,it calculates the similarity between words and words.Then we can get keywords by K-means algorithm.At the same time,this algorithm is better than TFIDF and LDA when 2-5 of 10 keywords,which is found in manual work.

关 键 词：word2vec 词向量 歌词关键词提取 K-MEANS 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]