[目的/意義]在社會網絡背景下,在線評論情感傾向已經成為影響觀影者決策的重要因素,如何有效提高在線評論情感分析的準確性成為學者們關注的熱點。[方法/過程]鑒于此,依據情感分析和概率語言術語集相關理論知識,提出一種新的影視推薦算法。首先運用TF-IDF算法提取主題詞,確定主題詞權重,然后計算在線評論情感值并細分區間確定情感程度,根據在線評論的情感程度構建概率語言決策矩陣,最后提出基于VIKOR的概率語言多準則決策方法,并將其用于電影排序。[結果/結論]采集Rotten Tomatoes官方網站上5部電影的真實在線評論數據,將文章提出的推薦算法與其他基于情感分析的推薦算法進行比較,驗證所提出算法的可行性和優越性。

[Purpose/significance] Under the background of social network,sentiment analysis of online reviews has become an important factor affecting the decision-making of movie viewers.How to effectively improve the efficiency and accuracy of sentiment analysis of online reviews has become a hot topic for scholars.[Method/process] In view of this,a new movie recommendation algorithm is proposed based on sentiment analysis and probabilistic linguistic term sets.Firstly,the Term Frequency-Inverse Document Frequency(TF-IDF) algorithm is used to extract the keywords and determine the weight of the keywords.Then the emotional value of online comments is calculated and the emotional degree is determined by dividing the intervals.the probabilistic linguistic decision matrix is constructed according to the emotional degree of online comments.Finally,a VIKOR method based on probabilistic linguistic term sets is proposed.[Result/conclusion] The real online comment data of five movies on Rotten Tomatoes official website are collected and compared with other recommendation algorithms based on sentiment analysis to verify the feasibility and superiority of the proposed algorithm.