[目的/意義]負面微博一般是指人們在微博平臺中表達用戶個人悲傷、恐懼、憤怒等情緒的微博,近幾年逐漸引起了人們的注意。高效快速地識別負面微博不僅可以幫助企業做產品改進、品牌營銷,同時還能幫助政府及時了解社會輿情等,因此對社會各領域都具有重要意義。[方法/過程]文章分析了負面微博有哪些特征,同時以健康領域作為案例,通過機器學習對提出的負面微博識別特征進行驗證。[結果/結論]實驗結果表明,不同領域負面微博所使用的負面詞都不一樣,同時也驗證了建立不同主題的微博負面詞表對于海量微博中進行負面微博的識別是有必要的,為人們做微博情感分析提供了新思路。[局限]文章選取的微博樣本量較少,選取機器學習的特征值以后可以添加句子語法特征等進行訓練。

[Purpose/significance] Negative sentiment microblog generally means that people express their personal sadness,fear,anger and other emotions in Weibo platform.In recent years,they have attracted people’s attention gradually.Identifying negative sentiment microblogs with efficiently and rapidly can not only help companies to do product improvement,brand marketing,but also help the government to understand social public opinion in a timely manner,so it is of great significance to all areas of society.[Method/process] The article analyses the features of negative tweets,and verify the proposed negative sentiment microblog identification characteristics through machine learning method.[Result/conclusion] The experimental results show that the different negative words used in different microblogs and areas and it is also necessary to construct the micro-blog negative thesaurus for identifying negative sentiment microblogs in different topics.It provides new ideas for people to do microblog sentiment analysis.[Limitations] The sample size of the microblog selected by the article is small.After selecting the features values of machine learning,sentence grammatical features could be added for training.