计算机科学
聚类分析
探索性研究
情报检索
树遍历
产品(数学)
探索性搜索
透视图(图形)
在线搜索
搜索引擎索引
万维网
机器学习
人工智能
数学
人类学
社会学
程序设计语言
几何学
标识
DOI:10.1016/j.ipm.2020.102323
摘要
With the fast growth of e-commerce and the emerging new retail trend—online and offline integration—it is important to recognize the target market and satisfy customers with different needs by analyzing their online search behaviors. Accordingly, we propose sequential search pattern analysis and clustering to analyze consumers' search behavior throughout the entire shopping process from the perspective of consumer need-states. We seek to understand how recommendation functions (RFs) or popular non-RF web features help consumers to shop online from a need-state perspective. We adopt maximal repeat patterns (MRPs) and lag sequential analysis (LSA) to analyze the sequence of search paths and identify significant repeated search patterns. Furthermore, to investigate the behaviors of customers with different types of need-states, we analyze webpages related to RFs and non-RF features using clustering to connect the evaluation results of search patterns with page traversal behaviors. This yields four groups of consumers who browse for information, adopt recommendations, consult reviews, and conduct searches with different levels of goal-oriented or exploratory-based need-states. The results show that consumers with strong goal-oriented need-states have the simplest search paths compared to other groups, whereas exploratory-based consumers have the most complicated search paths. Furthermore, consumers with higher need-states tend to search directly, consult reviews carefully, and have stored sequential search patterns, whereas consumers with exploratory-based need-states tend to explore the categories of products and adopt product classification hierarchy as a pivot to explore web features and then adopt specific types of RFs. Interestingly, consumers in the review-consulting group all belong to the goal-oriented need-states type with strong knowledge-building behaviors compared to others. The results reveal that each group employs its own particular web features to facilitate the shopping process and we can identify consumer types based on shopping behavior in the early stage of shopping. This suggests that e-store sellers can refine web features and deploy marketing strategies tailored to the search patterns for different levels of need-states.
科研通智能强力驱动
Strongly Powered by AbleSci AI