在线个性化产品设计的真实消费者需求识别研究

发布时间：2018-10-08 21:31

【摘要】：在线销售和在线个性化定制正以相对便捷和低价的特点,在家用、个人用品等行业快速兴起,并逐步被消费者、销售商、物流业和制造业界广泛接受。从消费者的角度看,在线历史数据,尤其是具有真实用户体验(Real User Experience,以下简称RUE)的在线评论不仅影响着后期消费者对产品性能的判断也影响着消费者的购买决策。从制造企业看,企业应该充分利用在线评论特别是基于RUE的在线评论,对其进行挖掘,识别实时的、真实的消费者需求,从而改进制造工艺、产品规格、营销和售后服务模式,形成新的更加个性化的产品,以便更好满足消费者的个性化需求。因此,商品在线评论信息的甄别、特征提取、消费倾向判定正成为企业关注的重点和学界研究的热点。然而在消费者需求识别过程中却存在以下三个问题；①越来越多的不具有RUE的消费者可以在Tmall网和TaoBao网上发布虚假的用户体验。②在特征词选择过程中,根据词语的某一特征选择产品特征词的方法无法归并原本意思相同或者相近的特征词,降低特征词数量。③针对多重否定评论中消费情感倾向的确定研究比较匮乏。综合以上考虑,本文提出了基于RUE的在线消费者需求获取模型、同质特征词识别和合并规则、多重否定评论中消费者情感判断方法等,并给出了应用实例。 1.具有RUE的在线评论获取研究。综合运用Web文本和Web网络日志信息,通过抽取Web文本信息,利用在线评论发布者的ID号对应的日志信息,选择日志信息中的浏览时间、浏览频率和评论次数等相关参数,编写能自动获取在线评论信息的应用程序,提出评论人是否具有真实用户体验的评判规则,剔除非RUE评论信息。 2.特征词的分类、合并和提取。由于中文评论中,表达同一意思可以用不同的字或者词语概括,汉字间的同义关系更加明显。因此,本文提出：在在线评论中抽取出名词或者名词词组作为候选特征词,通过判断符合同质特征词的条件,对同质特征词进行识别,利用合并规则将同质特征词归并,建立特征词映射库,通过映射,最终确定产品特征词。 3.多重否定评论的倾向判定。否定评论常包含着消费者对于商品改进及创新的期待,是制造企业进行产品创新的重要依据。针对网评中消费者常选择使用多个否定词来表达消费情感的特点,提出了多重否定评论情况下,判断消费者情感倾向的三个规则,提高对消费者需求识别的精度。 4.实证研究。以某品牌手机为例,运用本文提出的在线评论获取、特征词提取和消费倾向判定技术,实现了具有真实用户体验的网评鉴别、特征词选择和消费倾向判定,通过多个图形的对比,验证了本文提出方法和规则的可行性和优越性。
[Abstract]:Online sales and online personalized customization, with the characteristics of relative convenience and low price, has been widely accepted by consumers, distributors, logistics industry and manufacturing industry. From the consumer's point of view, online historical data, especially online reviews with real user experience (RUE), not only affect the consumer's judgment of product performance in the later period, but also affect consumer's purchase decision. From the point of view of manufacturing enterprises, enterprises should make full use of online reviews, especially those based on RUE, to mine them, identify real-time and real consumer needs, and thus improve manufacturing processes, product specifications, marketing and after-sales service models. Form new and more personalized products in order to better meet the personalized needs of consumers. Therefore, online commodity comment information screening, feature extraction, consumption tendency judgment is becoming the focus of attention and academic research focus. However, in the process of consumer demand identification, there are three problems: 1 more and more consumers who do not have RUE can publish false user experience .2 in the process of feature selection on Tmall and TaoBao. The method of selecting product feature words according to a certain feature of a word can not merge the original feature words with the same meaning or similar meaning. However, the research on reducing the number of feature words by 3. 3 in order to determine the tendency of consumer emotion in multiple negative comments is relatively scarce. Taking into account the above considerations, this paper proposes an online consumer demand acquisition model based on RUE, the recognition and merging rules of homogeneity features, and the method of consumer emotion judgment in multiple negative comments. Finally, an application example is given. 1. Research on online Review access with RUE. By synthetically using Web text and Web network log information, by extracting Web text information, using the corresponding log information of online comment publisher's ID number, selecting the related parameters such as browsing time, browsing frequency and comment times in the log information, etc. An application that can automatically obtain online comment information is written, and the rule of judging whether the reviewer has real user experience is proposed, and the non-RUE comment information is eliminated. 2. Classification, merging and extraction of feature words. The synonymy relationship between Chinese characters is more obvious because the same meaning can be summed up by different words or words in Chinese comments. Therefore, this paper proposes that nouns or noun phrases are extracted from online comments as candidate feature words, and homogenous feature words are identified by judging the condition of homogeneity feature words, and homogenous feature words are merged by merging rules. The mapping database of feature words is established, and the product feature words are finally determined by mapping. 3. The tendency of multiple negative comments to judge. Negative comments often contain consumers' expectations for product improvement and innovation, which is an important basis for manufacturing enterprises to carry out product innovation. In view of the characteristics that consumers often choose to use multiple negative words to express their consumer emotion in online evaluation, three rules are proposed to judge consumers' emotional tendency in the case of multiple negative comments to improve the accuracy of consumer demand recognition. 4. Empirical study. Taking a brand mobile phone as an example, using the techniques of online comment acquisition, feature word extraction and consumer propensity determination proposed in this paper, the online evaluation identification, feature word selection and consumer propensity determination with real user experience are realized. The feasibility and superiority of the proposed method and rules are verified by comparison of several graphs.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.1

【参考文献】