Customer Feedback Analytics from Unsolicited Resources

客户反馈管理（CFM）系统已经被广泛开发，以提高客户满意度和体验。然而，现有系统在很大程度上依赖于征求反馈渠道，例如调查，并且没有考虑在线社交平台，其中客户不受私人客户行业反馈渠道的约束，并且可以公开留下未经请求的反馈（社交反馈）。在这项提案中，我们希望成为第一批利用社交网络分析和自然语言处理技术来利用这些未经请求的公共反馈渠道的人。通过与Mitacs Accelerate支持合作的这个项目，我们的目标是：1）通过自动理解社会反馈的动态以及如何反映特定客户的需求和偏好（客户服务）来提高客户满意度; 2）提高客户保留率，增加商业机会并识别紧急社区趋势（业务支持）。我们在这项提案中的工业合作伙伴Press'nXPress是世界上第一家也是唯一一家将社交渠道与物理接触点连接起来的加拿大公司。最重要的成果将是下一代具有社会意识的CFM，它将被越来越多的社会反馈所采用，从而提高加拿大作为CFM解决方案世界领导者之一的地位。其次，加拿大各行业将利用这一项目的成果改善其客户服务，保留现有客户，寻找新的潜在客户，并赢回以前的客户。从学术角度来看，在为公平和包容性的研究和工业环境中培养高素质人才提供坚实基础的同时，该提案将弥合社交媒体分析和客户评论分析研究。与现有的基于图的社交内容处理技术不同，我们的工作将集中在使用语言建模和神经网络技术来理解社交反馈。我们将1）对服务的方面及其时间特征建模，2）识别以相关服务为中心的社区，3）检测基于方面的紧急趋势。

Customer Feedback Management (CFM) systems have been extensively developed for improved client satisfaction and experience. Existing systems, however, have relied much on solicited feedback channels, such as surveys and have not considered online social platforms where the customers are not constrained in private customer-industry feedback channels and can leave unsolicited feedback publicly (social feedback). In this proposal, we seek to be among the first to leverage these unsolicited public feedback outlets using social network analysis and natural language processing techniques. Through this project jointly with Mitacs Accelerate support, we aim at 1) enhancing customer satisfaction by automatically understanding the dynamics of the social feedback and how that reflects on specific customer's needs and preferences (customer care); and 2) increasing customer retention, enhancing business opportunities and identifying emergent community trends (business support). Our industrial partner in this proposal, Press'nXPress, is the world's first and only Canadian company that will connect social channels with physical touchpoints. The foremost outcome will be the next-generation social-aware CFM that will be adopted by the growing social feedback, and therefore, enhance Canada's position as one of the world leaders in CFM solutions. Secondly, Canadian industries will leverage the outcomes of this project to improve their customer care, yielding to the retention of current customers, finding new prospective customers, and winning back former customers. From an academic perspective, while providing a solid foundation for the training of highly qualified personnel in equitable and inclusive research and industrial environment, this proposal will bridge social media analytics and customer review analysis research. Unlike the existing graph-based techniques for processing social content, our work will center on the use of language modelling and neural network technologies to make sense of social feedback. We will 1) model services' aspects and their temporal characteristics, 2) identify communities that are centred around related services, and 3) detect emergent aspect-based trends.