餐饮服务评价情感倾向分析——基于不同分类模型的比较  

Analysis of Affective Tendency of Catering Service Evaluation—Comparison Based on Different Classification Models

作　　者：蔡翔宇 

机构地区：[1]成都信息工程大学,应用数学学院,四川 成都

出　　处：《应用数学进展》2023年第3期940-952,共13页Advances in Applied Mathematics

摘　　要：餐饮业作为“永不落幕的黄金行业”,近年受网络发展与疫情的影响,呈现“堂食”与“外卖”并存之势。为提高餐饮业的竞争力,准确把握用户的需求倾向尤为重要,本文对不同餐饮企业的餐品评论数据进行数据分析,构建不同的分类模型对餐饮服务评论的情感倾向进行分析,并通过计算精确率、召回率以及F1-Score对模型性能和误差进行评估。分别构建K近邻分类模型、LSTM长短时记忆模型、BiLSTM模型,以及CNN-Multi-BiLSTM模型对情感倾向进行分类;结果表明CNN-Multi-BiLSTM模型具有较高准确性,平均精确率、召回率和F1-Socre分别达到了91.5%、91.35%,91.45%。因此,将CNN-Multi-BiLSTM模型用来评测餐饮服务评论数据,可让商家更加精准地把握用户需求,制定相应改进策略,提高其竞争力。The catering industry, as the “never ending gold industry”, has been affected by the development of the Internet and the epidemic in recent years, showing the coexistence of “sit-down food” and “take-out” trend. In order to improve the competitiveness of the catering industry, it is particularly important to accurately grasp the demand tendency of users. In this paper, data analysis was con-ducted on the food review data of different catering enterprises, different classification models were constructed to analyze the emotional tendency of food service reviews, and the model performance and error were evaluated by calculating the accuracy rate, recall rate and F1-Score. The K-nearest neighbor classification model, LSTM short and short term memory model, BiLSTM model, and CNN-Multi-BiLSTM model were constructed respectively to classify the emotional tendency. The re-sults show that the CNN-Multi-BiLSTM model has high accuracy, with the average accuracy rate, re-call rate and F1-Socre reaching 91.5%, 91.35% and 91.45%, respectively. Therefore, the CNN-Multi-BiLSTM model can be used to evaluate the data of catering service reviews, so that busi-nesses can more accurately grasp the needs of users, formulate corresponding improvement strat-egies and improve their competitiveness.

关 键 词：情感倾向分析 数据分析 K近邻 LSTM模型 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]