Multi-modal Characteristics Analysis and Customer Service Efficiency Improvement in the Digital Community Based on User Clustering

Abstract

[Purpose/Significance] Multi-modal feature analysis and service efficiency improvement of digital community consumers will help to provide a new vision for the construction of digital intelligent online communities and provide new impetus for relevant departments to make decisions. In addition, although the current research on digital consumption includes the relevant content of user analysis, it mainly aims at the formulation of detailed operation plans, and lacks the analysis of service efficiency improvement of digital communities. On the other hand, the research on user value orientation for online service quality optimization is mostly based on profile technology, which only considers the difference characteristics of a single target user, and lacks the horizontal comparison and difference attribution research of multi-modal features among groups. Based on this, this paper, from the perspective of value discovery, achieves clustering by aggregating user profiles, analyzes the multi-modal characteristics of consumer groups in digital communities, and proposes a service efficiency improvement scheme. [Method/process] First, this paper analyzed the target consumers in the digital community and established a cluster indicator system. Then, users were grouped, and the multi-modal information profile of the target group was restored based on group characteristics and inter-group interaction characteristics. Finally, it proposed the path to improve the efficiency of digital community services. In terms of technical implementation, the data related to consumer activities were extracted from the digital community, integrated, cleaned, and distributed to the storage bucket. The clustering indicator system was built through feature mining and existing indicators, and the indicators were mapped to aims, and DBSCAN clustering was carried out on the basis of using AP to realize the image. After grouping and naming, the characteristics analysis, interaction analysis, and drift and penetration phenomenon analysis were carried out according to the characteristics of various groups. We extracted various parameters of the design of digital community consumption activities, and built a decision variable function to find the optimal behavior equilibrium conditions of the digital product supplier, consumer and digital community. Based on this, we built an efficiency improvement tree, and proposed community service efficiency improvement strategies at the initial, middle and later stages of consumption activities. [Results/Conclusions] The empirical analysis results show that the model in this paper can first generate reasonable and effective clustering results, and then realize the classification of group characteristics and the analysis of inter-group infiltration and drift. The clustering results show six types of consumer groups: focus, center, special, sleeping, loss and general groups. Most groups will have user penetration, and only general user groups will have inter-group drift. The service efficiency improvement model shows that the most valued group is the center and key group. The inadequacy of this study is that the applicability of the model to multi-source heterogeneous data needs to be tested and there is still room for improvement in clustering granularity.

Keywords

• user clustering|ap-dbscan|multi-modal characteristics|digital community|digital consumption