Abstract: In the era of mobile Internet, a large number of APP users pay more attention to product experience and express their usage and suggestions through comments. Research on online comment data has become a hot topic, and user feedback from comments is conducive to APP evolution and upgrading. But comment mining for APP is in the ascendant. In this paper, a large number of user comment data are collected from 9 APP stores to screen the demand attributes and emotional tendencies contained in the comments. The KANO model is used for modeling and analysis, and the attributes are mapped to attractive quality, one-dimensional quality, must-be quality and other cate-gories. According to the KANO category of the APP attribute, a reasonable and effective update evolution strategy is proposed: the APP developers should give priority to meeting the requirements of one-dimensional quality and must-be quality, and gradually realize the requirements of attractive quality. Finally, this paper proves the robustness and portability of the method.

Key words: crowdsourced data, application (APP) user reviews, KANO model, demand measurement, evolution strategy

摘要： 移动互联网时代中，APP用户更注重产品体验，通过评论的方式来表达自己的使用情况和建议。在线评价数据的研究已经成为热点，从评论中获得的用户反馈有助于APP演化升级，但目前针对APP的评论挖掘方兴未艾。从9家APP应用商店中采集得到大量用户评论数据，筛选评论所包含的需求属性和情感倾向，并运用KANO模型对其建模分析，映射属性到魅力、期望、必备等类别。根据APP具体属性和所属KANO类别给出合理有效的更新演化策略：APP演化应优先满足必备和期望属性的需求，并逐步实现魅力属性的需求，并且最终检验了模型的鲁棒性和易移植性。

关键词: 群智数据, APP用户评论, KANO模型, 需求计量, 演化策略