网络商品评论在线监控.pptx

Monitoring of User-Generated Reviews via a Sequential Reverse Joint Sentiment-Topic ModelOutlineBackground and introductionMethodologySequential reverse joint sentiment-topic (SRJST) modelSRJST chartAlternative approaches: RJST and RJST-EWMA chartsCase studySimulation studyConclusionBackground and introductionBooming e-commerce in China网购用户规模5.33亿快递服务业业务累计量400.6亿件非银行支付机构发生网络支付金额143.26万亿元Background and introductionOnline user-generated reviews (UGR)An efficient tool for evaluating the customer-perceived quality of online products and services.These reviews supply two types of hidden information for evaluating the quality of online products and services: the topics that correspond to quality characteristics, and the sentiments (or customer opinions) conditioned on specific topics.SentimentsTopics PositiveNegativeAppearanceFunctionDelivery …Online product reviewsBackground and introductionOnline user-generated reviews (UGR)Extract meaningful features and retrieve customer opinions in the field of information retrieval (IR) Statistical process control (SPC)Monitor various quality characteristics in many cases, including continuous and categorical processesSPC can be applied to online UGR for quickly detecting the hidden evolution in customer opinions and monitoring the quality of products and relevant services in an after-sales stageMethodologyTargetMonitor the quantitative evolution of both topics and sentiments in UGRTwo stepsWe propose a sequential reverse joint sentiment-topic (SRJST) model to quantify the discrete documents along the time lineBased on the results of SRJST model, a joint monitoring scheme (SRJST chart) is proposed for monitoring the quantitative evolution of both topics and sentimentsSRJST model?WordA basic unit from a vocabulary indexed by DocumentA collection of words denoted by where the the sequence of words is exchangeable (bag-of-words assumption)For example, a document is a collection of all words gathered from reviews in one day (for daily