本文選題：評分預(yù)測 + 情感分析��； 參考：《數(shù)據(jù)分析與知識發(fā)現(xiàn)》2017年08期

【摘要】：【目的】通過網(wǎng)絡(luò)用戶評論,為評論網(wǎng)站構(gòu)建有效的評分預(yù)測機制。【方法】提出基于網(wǎng)絡(luò)用戶評論的評分預(yù)測模型,該模型包括4個模塊:網(wǎng)絡(luò)用戶評論獲取模塊、預(yù)測變量獲取模塊、預(yù)測分析模塊以及預(yù)測結(jié)果評價模塊。抓取30部不同類型的電影評論數(shù)據(jù),27部用于構(gòu)建模型,3部用于檢驗?zāi)Ｐ��！窘Y(jié)果】使用逐步回歸方法篩選出變量:參與評分人數(shù)、參與評論人數(shù)、想要觀看人數(shù)和電影正向評論情感均值,構(gòu)建評分預(yù)測模型。使用3部電影驗證,預(yù)測評分與IMDb評分相差最大值為0.0644,最小值為0.0227�！揪窒蕖吭跀�(shù)據(jù)樣本量、情感特征提取精度、模型普適性驗證等方面有待進一步提升�！窘Y(jié)論】該模型能夠依據(jù)用戶評論對評分進行有效預(yù)測,在網(wǎng)絡(luò)水軍探測方面也能發(fā)揮一定的作用。
[Abstract]:[objective] to construct an effective scoring prediction mechanism for the comment website through the network user comments. [methods] A scoring prediction model based on the network user comment is proposed. The model includes four modules: the network user comment acquisition module. Prediction variable acquisition module, prediction analysis module and prediction evaluation module. Grabbing 30 different types of movie review data, 27 were used to build models and 3 were used to test the models. [results] the variables were screened out using stepwise regression: number of participants, number of comments, In order to evaluate the number of viewers and the emotional average of positive reviews, a rating prediction model was constructed. Using three films to verify, the difference between prediction score and IMDb score is 0.0644, the minimum value is 0.0227.The accuracy of data sample size, emotion feature extraction, [conclusion] the model can effectively predict the score according to the user comments, and it can also play a certain role in the detection of the network navy.
【作者單位】： 中山大學資訊管理學院;
【基金】：國家社會科學基金項目“用戶評論情感分析及其在競爭情報服務(wù)中的應(yīng)用研究”(項目編號:11CTQ022) 廣東省科技專項“基于內(nèi)容的科技文獻分析服務(wù)平臺”(項目編號:2016B030303003)的研究成果之一
【分類號】：G252
，

本文編號：1818518