近年来,互联网蓬勃发展,各种产品的使用用户会在互联网上发表自己的售后评价,这些评价中蕴含着很多潜在的信息,特别是文字评价。因此,本文首先采用了基于词典的情感分析将文字评价转换为情感得分,再进行后续的研究。情感得分代表着用户...近年来,互联网蓬勃发展,各种产品的使用用户会在互联网上发表自己的售后评价,这些评价中蕴含着很多潜在的信息,特别是文字评价。因此,本文首先采用了基于词典的情感分析将文字评价转换为情感得分,再进行后续的研究。情感得分代表着用户对产品及服务的态度,当情感得分发生异常变化时,意味着该时段产品或服务有着质量下降的趋势,因此,对用户的在线评论进行监控有着极其重要的意义。当情感得分失控时,我们能及时告知产品经理,以便采取一些相关措施,减少不必要的损失。在此基础上,本文提出了非参数的Cucconi-EWMA控制图,来对情感得分的位移参数和尺度参数同时进行监控。最后,将本文提出的非参数控制图用于了实际案例上,监控效果良好,进一步证实了本文控制图对在线评论监控的有效性。In recent years, the Internet has developed rapidly, and users of various products will publish their after-sales reviews on the Internet. These reviews contain a lot of potential information, especially text reviews. Therefore, this paper first uses dictionary-based sentiment analysis to convert text reviews into sentiment scores, and then conducts subsequent research. Sentiment scores represent the attitude of users towards products and services. When the sentiment score changes abnormally, it means that the product or service has a trend of declining quality during this period. Therefore, it is extremely important to monitor users’ online reviews. When the sentiment score is out of control, we can inform the product manager in time so that some relevant measures can be taken to reduce unnecessary losses. On this basis, this paper proposes a non-parametric Cucconi-EWMA control chart to monitor the displacement parameters and scale parameters of the sentiment score at the same time. Finally, the non-parametric control chart proposed in this paper is used in actual cases, and the monitoring effect is good, which further confirms the effectiveness of the control chart in this paper for online review monitoring.展开更多
文摘近年来,互联网蓬勃发展,各种产品的使用用户会在互联网上发表自己的售后评价,这些评价中蕴含着很多潜在的信息,特别是文字评价。因此,本文首先采用了基于词典的情感分析将文字评价转换为情感得分,再进行后续的研究。情感得分代表着用户对产品及服务的态度,当情感得分发生异常变化时,意味着该时段产品或服务有着质量下降的趋势,因此,对用户的在线评论进行监控有着极其重要的意义。当情感得分失控时,我们能及时告知产品经理,以便采取一些相关措施,减少不必要的损失。在此基础上,本文提出了非参数的Cucconi-EWMA控制图,来对情感得分的位移参数和尺度参数同时进行监控。最后,将本文提出的非参数控制图用于了实际案例上,监控效果良好,进一步证实了本文控制图对在线评论监控的有效性。In recent years, the Internet has developed rapidly, and users of various products will publish their after-sales reviews on the Internet. These reviews contain a lot of potential information, especially text reviews. Therefore, this paper first uses dictionary-based sentiment analysis to convert text reviews into sentiment scores, and then conducts subsequent research. Sentiment scores represent the attitude of users towards products and services. When the sentiment score changes abnormally, it means that the product or service has a trend of declining quality during this period. Therefore, it is extremely important to monitor users’ online reviews. When the sentiment score is out of control, we can inform the product manager in time so that some relevant measures can be taken to reduce unnecessary losses. On this basis, this paper proposes a non-parametric Cucconi-EWMA control chart to monitor the displacement parameters and scale parameters of the sentiment score at the same time. Finally, the non-parametric control chart proposed in this paper is used in actual cases, and the monitoring effect is good, which further confirms the effectiveness of the control chart in this paper for online review monitoring.
