中文微博情感分類的研究
發(fā)布時間：2019-06-22 19:08
【摘要】：隨著互聯(lián)網(wǎng)的發(fā)展,在Web2.0時代的主流網(wǎng)絡(luò)社交平臺中,微博已經(jīng)是被廣大互聯(lián)網(wǎng)用戶使用最頻繁的社交工具。它因為書寫簡短,發(fā)布方便,以及能進(jìn)行實時互動等諸多特點受到大眾的歡迎。用戶逐漸傾向使用微博工具來向外界分享自己的內(nèi)容,進(jìn)而來表達(dá)自己的觀點、看法和情緒。微博影響力的日益擴大,也吸引了大批研究者的關(guān)注,其中對微博進(jìn)行情感分類就是相關(guān)領(lǐng)域中一個重要的研究方向。當(dāng)前關(guān)于英文微博的情感分類的研究比較充分,而有關(guān)中文微博的情感分類研究還處于初始階段。中文微博用戶日益增多,微博已經(jīng)開始影響國人的方方面面,所以開展中文微博的情感分類研究顯得非常重要和緊迫。 情感分類研究主要是通過分析和挖掘文本中帶有情感性的主觀性內(nèi)容,以此對文本的情感所屬類別做出判斷。本文將分析中文微博本身具有的特征,在傳統(tǒng)文本情感分類已有相關(guān)理論和方法上,對中文微博的情感分類進(jìn)行研究。在中文微博文本的主客觀分類研究中,論文提出一種基于詞典與語料結(jié)合的中文微博主觀句抽取方法,首先通過一個高可信的情感詞典抽取句子中的情感表達(dá)文本,以保證結(jié)果的準(zhǔn)確率；而后基于句子2-POSW模型通過語料學(xué)習(xí)的方法抽取句子中的剩余情感表達(dá)文本,從而提高了召回率。在中文微博文本的情感極性分類研究中,論文首先抽取出中文微博中的主觀句部分,然后參考微博表情標(biāo)注結(jié)果和高可信情感詞典標(biāo)注結(jié)果,構(gòu)建了中文微博的情感極性語料庫,在保證了語料庫規(guī)模的同時確保了標(biāo)注的質(zhì)量,并減輕了人工標(biāo)注的負(fù)擔(dān)。在建立的微博情感極性語料庫的基礎(chǔ)上,抽取情感詞和情感短語特征,并利用頻度和信息熵進(jìn)行優(yōu)化,并結(jié)合情感詞典特征以及標(biāo)點符號特征,進(jìn)行了中文微博情感極性分類的實驗。實驗結(jié)果表明,在中文微博的主客觀分類中,相比于傳統(tǒng)的基于大規(guī)模情感詞典的方法,本文方法的F值提高了7%。在中文微博的情感極性分類中,本文提出的優(yōu)化方法的F值可以達(dá)到81.9918%,取得了較好的實驗結(jié)果。
[Abstract]:With the development of the Internet, Weibo has been the most frequently used social tool by the majority of Internet users in the mainstream social platform of the Web2.0 era. It is welcomed by the public because of its short writing, convenient release, and real-time interaction. Users tend to use Weibo tools to share their content with the outside world, and then to express their views, views and emotions. The increasing influence of Weibo has also attracted the attention of a large number of researchers, among which the emotional classification of Weibo is an important research direction in related fields. At present, the research on emotional classification of English Weibo is relatively sufficient, but the research on emotional classification of Chinese Weibo is still in its initial stage. With the increasing number of Chinese Weibo users, Weibo has begun to affect all aspects of the Chinese people, so it is very important and urgent to carry out the research on emotional classification of Chinese Weibo. The research of emotion classification mainly analyzes and excavates the subjective content with emotion in the text, so as to judge the category of emotion in the text. This paper will analyze the characteristics of Chinese Weibo itself, and study the emotional classification of Chinese Weibo in the existing theories and methods of traditional text emotion classification. In the subjective and objective classification of Chinese Weibo text, this paper proposes a Chinese Weibo subjective sentence extraction method based on dictionary and corpus. Firstly, the emotional expression text in the sentence is extracted through a highly credible emotional dictionary to ensure the accuracy of the result; then, based on the sentence 2-POSW model, the remaining emotional expression text in the sentence is extracted by corpus learning, thus improving the recall rate. In the research of emotional polarity classification of Chinese Weibo text, firstly, the subjective sentence part of Chinese Weibo is extracted, and then the emotional polarity corpus of Chinese Weibo is constructed by referring to the results of Weibo expression tagging and highly trusted emotion dictionary, which not only ensures the scale of corpus, but also ensures the quality of tagging and lightens the burden of manual tagging. Based on the Weibo emotional polarity corpus, the emotional words and emotional phrase features are extracted, and the frequency and information entropy are used to optimize the emotional polarity classification. Combined with the emotional dictionary features and punctuation symbol features, the experiment of Chinese Weibo emotional polarity classification is carried out. The experimental results show that the F value of this method is 7% higher than that of the traditional method based on large-scale emotion dictionary in the subjective and objective classification of Chinese Weibo. In the emotional polarity classification of Chinese Weibo, the F value of the optimization method proposed in this paper can reach 81.9918%, and good experimental results are obtained.
【學(xué)位授予單位】：華東師范大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP391.1;TP393.092
[Abstract]:With the development of the Internet, Weibo has been the most frequently used social tool by the majority of Internet users in the mainstream social platform of the Web2.0 era. It is welcomed by the public because of its short writing, convenient release, and real-time interaction. Users tend to use Weibo tools to share their content with the outside world, and then to express their views, views and emotions. The increasing influence of Weibo has also attracted the attention of a large number of researchers, among which the emotional classification of Weibo is an important research direction in related fields. At present, the research on emotional classification of English Weibo is relatively sufficient, but the research on emotional classification of Chinese Weibo is still in its initial stage. With the increasing number of Chinese Weibo users, Weibo has begun to affect all aspects of the Chinese people, so it is very important and urgent to carry out the research on emotional classification of Chinese Weibo. The research of emotion classification mainly analyzes and excavates the subjective content with emotion in the text, so as to judge the category of emotion in the text. This paper will analyze the characteristics of Chinese Weibo itself, and study the emotional classification of Chinese Weibo in the existing theories and methods of traditional text emotion classification. In the subjective and objective classification of Chinese Weibo text, this paper proposes a Chinese Weibo subjective sentence extraction method based on dictionary and corpus. Firstly, the emotional expression text in the sentence is extracted through a highly credible emotional dictionary to ensure the accuracy of the result; then, based on the sentence 2-POSW model, the remaining emotional expression text in the sentence is extracted by corpus learning, thus improving the recall rate. In the research of emotional polarity classification of Chinese Weibo text, firstly, the subjective sentence part of Chinese Weibo is extracted, and then the emotional polarity corpus of Chinese Weibo is constructed by referring to the results of Weibo expression tagging and highly trusted emotion dictionary, which not only ensures the scale of corpus, but also ensures the quality of tagging and lightens the burden of manual tagging. Based on the Weibo emotional polarity corpus, the emotional words and emotional phrase features are extracted, and the frequency and information entropy are used to optimize the emotional polarity classification. Combined with the emotional dictionary features and punctuation symbol features, the experiment of Chinese Weibo emotional polarity classification is carried out. The experimental results show that the F value of this method is 7% higher than that of the traditional method based on large-scale emotion dictionary in the subjective and objective classification of Chinese Weibo. In the emotional polarity classification of Chinese Weibo, the F value of the optimization method proposed in this paper can reach 81.9918%, and good experimental results are obtained.
【學(xué)位授予單位】：華東師范大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP391.1;TP393.092
【相似文獻(xiàn)】
相關(guān)碩士學(xué)位論文 前3條
1 朱海歡;中文微博情感分類的研究[D];華東師范大學(xué);2014年
2 林江豪;中文微博情感分析關(guān)鍵技術(shù)研究[D];廣東外語外貿(mào)大學(xué);2013年
3 彭蔚U，
本文編號：2504882
本文鏈接：http://sikaile.net/guanlilunwen/ydhl/2504882.html
最近更新
教材專著
