基于主动学习的新媒体政务互动内容情感挖掘研究

Research on Emotion Mining of Government Affairs Interactive Content in New Media Based on Active Learning

[目的/意义]新媒体平台逐渐成为政民交互的重要载体,准确把握新媒体政务互动内容中的情感倾向,有助于提升政府舆情把握能力与社会治理能力。[方法/过程]在BERT文本语义表示基础上,将主动学习策略与BiLSTM模型集成,进行新媒体政务互动内容情感倾向分析,以提升模型对互动内容情感数据的有效利用。[结果/结论]针对“法律法规草案公开征求意见类”微博互动内容的实验表明,将主动学习引入BERT-BiLSTM模型后,模型的准确率、召回率及F值提升,新媒体政务互动内容情感呈现效果较好。文章所提模型科学可行,能够在减少数据依赖的情况下,提升情感挖掘的效率。

[Purpose/significance] New media platform has gradually become an important carrier of the interaction between the government and the people.Accurately grasping the emotional tendency of government affairs interactive content in new media helps to improve the government’s ability of public opinion grasping and social governance.[Method/process] This paper integrates active learning strategy with BiLSTM model on the basis of BERT text semantic representation for sentiment tendency analysis of government affairs interactive content,in order to improve the effective use of interactive content sentiment data.[Result/conclusion] Experiment on government affairs interactive content of “draft laws and regulations for public consultation” in Sina Microblog shows that after active learning is introduced into BERT-BilSTM model,the accuracy rate,recall rate and F value of the model are improved,and the emotion of government affairs interactive content can be better presented.The proposed model in this paper is scientifically feasible and can improve the efficiency of sentiment mining with less data dependence.