微博信息分類研究
發(fā)布時間：2018-11-21 21:23
【摘要】：微博作為一種新興的網(wǎng)絡(luò)交互平臺,包含著海量的信息。在這些以微博為載體的信息中蘊(yùn)含著巨大的潛在的商機(jī)。微博作為社交網(wǎng)絡(luò)平臺,需要企業(yè)對其產(chǎn)品的使用者或者需求者做出及時的服務(wù)。對于微博內(nèi)容中流露出的對企業(yè)產(chǎn)品的抱怨或者針對產(chǎn)品的疑問,企業(yè)有義務(wù)也有責(zé)任在第一時間及時的對產(chǎn)品使用者提供相關(guān)的技術(shù)支持服務(wù)；對于微博內(nèi)容中表現(xiàn)出對企業(yè)產(chǎn)品的求購信息,企業(yè)也需要在第一時間及時的提供產(chǎn)品的相關(guān)信息,并做出相應(yīng)的導(dǎo)購服務(wù)。 任何企業(yè),面對海量的微博信息,如何及時的從數(shù)以億計的微博信息中挖掘出企業(yè)需要的微博內(nèi)容是當(dāng)代企業(yè)獲得第一手信息的關(guān)鍵。在獲得與企業(yè)及其產(chǎn)品相關(guān)的微博內(nèi)容的同時,對相關(guān)聯(lián)的信息進(jìn)行準(zhǔn)確的情感傾向分類和做出是否需要提供導(dǎo)購服務(wù)或者產(chǎn)品技術(shù)支持服務(wù)是企業(yè)需要關(guān)注的一項重要任務(wù)。 本文將以聯(lián)想集團(tuán)作為企業(yè)背景,將微博文本根據(jù)是否與聯(lián)想集團(tuán)及其產(chǎn)品相關(guān)、微博文本中流露出的情感傾向性、是否需要對用戶提供導(dǎo)購服務(wù)、是否需要對用戶提供技術(shù)支持服務(wù)對微博進(jìn)行分類。根據(jù)微博信息分類系統(tǒng)的功能性要求,設(shè)計系統(tǒng)的總體架構(gòu)。根據(jù)微博信息分類系統(tǒng)需要實(shí)現(xiàn)的不同分類功能,設(shè)計并實(shí)現(xiàn)分類器,著重設(shè)計分類器的特征提取算法,并與簡單特征提取算法進(jìn)行對比試驗(yàn)。為了檢測微博信息分類系統(tǒng)的功能及分類器的分類效果,將對分類器進(jìn)行測試,以期到達(dá)可以進(jìn)行微博信息分類的理想結(jié)果。 本文的最后部分,將對微博信息分類系統(tǒng)的分類器測試結(jié)果進(jìn)行分析。在本文分類器使用的特征提取算法和簡單特征提取算法進(jìn)行對比試驗(yàn)的結(jié)果分析中,得出本文中使用的特征提取算法分類效果更好。在單獨(dú)對分類器的分類效果進(jìn)行分析時,會著重考慮特征的查全率、查準(zhǔn)率和調(diào)和值,分析分類器的具體分類效果。在分析分類效果之后,會根據(jù)測試結(jié)果分析微博信息分類系統(tǒng)目前的不足和待改進(jìn)的地方,并提出微博信息分類系統(tǒng)進(jìn)一步的改進(jìn)方案。最后,本文將是對微博信息分類系統(tǒng)的展望。
[Abstract]:Weibo, as a new network interactive platform, contains a great deal of information. In these Weibo as the carrier of information contains a huge potential business opportunities. Weibo, as a social network platform, needs enterprises to provide timely service to the users or demanders of their products. As for the complaints or questions about the products of the enterprises revealed in Weibo's contents, the enterprises have the obligation and the responsibility to provide the relevant technical support services to the users of the products in the first time. For Weibo's content to show the purchase information of the enterprise's products, the enterprise also needs to provide the relevant information of the product in the first time, and make the corresponding purchasing guide service. Faced with the massive Weibo information, it is the key for contemporary enterprises to obtain first-hand information from hundreds of millions of Weibo information. While obtaining Weibo's content related to enterprises and their products, it is an important task for enterprises to classify the relevant information accurately and to make whether or not they need to provide purchasing guidance services or product technical support services. This article will take the Lenovo Group as the enterprise background, according to whether Weibo text is related to the Lenovo Group and its products, and whether the emotional tendency revealed in Weibo's text, whether or not it needs to provide the purchasing guide service to the users. Whether need to provide the technical support service to the user to carry on the classification to Weibo. According to the functional requirements of Weibo information classification system, the overall structure of the system is designed. According to the different classification functions of Weibo information classification system, the classifier is designed and implemented. The feature extraction algorithm of the classifier is designed and compared with the simple feature extraction algorithm. In order to test the function of Weibo information classification system and the classification effect of the classifier, the classifier will be tested in order to reach the ideal result which can be used for the information classification of Weibo. In the last part of this paper, the test results of Weibo information classification system are analyzed. In the analysis of the result of comparison between the feature extraction algorithm and the simple feature extraction algorithm used in this paper, it is concluded that the feature extraction algorithm used in this paper has better classification effect. In the analysis of the classification effect of the classifier separately, the recall rate, precision rate and harmonic value of the feature will be considered emphatically, and the concrete classification effect of the classifier will be analyzed. After analyzing the classification effect, the paper will analyze the deficiency of Weibo information classification system and the place that need to be improved according to the test results, and put forward the further improvement scheme of Weibo information classification system. Finally, this paper will be the prospect of Weibo information classification system.
【學(xué)位授予單位】：山東大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP393.092;TP391.1
本文編號：2348289
[Abstract]:Weibo, as a new network interactive platform, contains a great deal of information. In these Weibo as the carrier of information contains a huge potential business opportunities. Weibo, as a social network platform, needs enterprises to provide timely service to the users or demanders of their products. As for the complaints or questions about the products of the enterprises revealed in Weibo's contents, the enterprises have the obligation and the responsibility to provide the relevant technical support services to the users of the products in the first time. For Weibo's content to show the purchase information of the enterprise's products, the enterprise also needs to provide the relevant information of the product in the first time, and make the corresponding purchasing guide service. Faced with the massive Weibo information, it is the key for contemporary enterprises to obtain first-hand information from hundreds of millions of Weibo information. While obtaining Weibo's content related to enterprises and their products, it is an important task for enterprises to classify the relevant information accurately and to make whether or not they need to provide purchasing guidance services or product technical support services. This article will take the Lenovo Group as the enterprise background, according to whether Weibo text is related to the Lenovo Group and its products, and whether the emotional tendency revealed in Weibo's text, whether or not it needs to provide the purchasing guide service to the users. Whether need to provide the technical support service to the user to carry on the classification to Weibo. According to the functional requirements of Weibo information classification system, the overall structure of the system is designed. According to the different classification functions of Weibo information classification system, the classifier is designed and implemented. The feature extraction algorithm of the classifier is designed and compared with the simple feature extraction algorithm. In order to test the function of Weibo information classification system and the classification effect of the classifier, the classifier will be tested in order to reach the ideal result which can be used for the information classification of Weibo. In the last part of this paper, the test results of Weibo information classification system are analyzed. In the analysis of the result of comparison between the feature extraction algorithm and the simple feature extraction algorithm used in this paper, it is concluded that the feature extraction algorithm used in this paper has better classification effect. In the analysis of the classification effect of the classifier separately, the recall rate, precision rate and harmonic value of the feature will be considered emphatically, and the concrete classification effect of the classifier will be analyzed. After analyzing the classification effect, the paper will analyze the deficiency of Weibo information classification system and the place that need to be improved according to the test results, and put forward the further improvement scheme of Weibo information classification system. Finally, this paper will be the prospect of Weibo information classification system.
【學(xué)位授予單位】：山東大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP393.092;TP391.1
【參考文獻(xiàn)】
相關(guān)期刊論文 前3條
1 丁國棟;白碩;王斌;;文本檢索的統(tǒng)計語言建模方法綜述[J];計算機(jī)研究與發(fā)展;2006年05期
2 李衛(wèi)疆;趙鐵軍;王憲剛;;基于上下文的查詢擴(kuò)展[J];計算機(jī)研究與發(fā)展;2010年02期
3 朱嫣嵐;閔錦;周雅倩;黃萱菁;吳立德;;基于HowNet的詞匯語義傾向計算[J];中文信息學(xué)報;2006年01期
，本文編號：2348289
本文鏈接：http://sikaile.net/guanlilunwen/ydhl/2348289.html
最近更新
教材專著
