基于LDA的公共文化主题提取与演化分析

Extraction and Evolution Analysis of Public Culture Topic Based on LDA Model

[目的/意义]当前公共文化活动丰富多彩、各种新举措层出不穷,如何快速全面地了解公共文化发展的各类模式、最新动态,成为公共文化领域需要解决的一个问题。[方法/过程]采用scrapy爬虫获取各省级图书馆、文化馆官网所发布的活动资讯和新闻报道文本,经过预处理后对文本进行分词并使用TF-IDF算法提取关键词,结合LDA主题模型进行主题建模并分类。根据主题分类结果对所有文档进行主题标注,从时间和空间维度对文档主题进行统计分析和可视化展示。[结果/结论]从时间上看,个别主题近年来逐渐成为热点,相关新闻活动数量迅速增长;从省域分布角度来看,各省所反映的主题具有较强的差异性。

[Purpose/significance]Current public cultural activities are rich and colorful,and various new measures emerge one after another.How to quickly and comprehensively understand various modes and latest trends of public cultural development has become a problem demanding future solution in the field of public culture.[Method/process]In order to solve such problems,this paper adopts web crawler scrapy to obtain activity information and news report texts published on the official websites of provincial libraries and cultural centers.After preprocessing,According to the results of subject classification,all documents are labelled with subjects,and then the document subjects are statistically analyzed and visualized from the dimensions of time and space.[Result/conclusion]Some topics have recently become hotspots in terms of time,and the number of relevant news reporting activities has increased rapidly.From the perspective of provincial distribution,there are strong differences between the topics reflected by each province.