判断
计算机科学
潜在Dirichlet分配
偏爱
分析
产品(数学)
数据科学
启发式
旅游
领域(数学)
主题模型
运筹学
功能(生物学)
语句(逻辑)
问题陈述
任务(项目管理)
情绪分析
机器学习
情报检索
人工智能
推荐系统
数据挖掘
消费者行为
许可
电话
知识管理
作者
Yu Yang,Yelin Fu,Y J Zhang
摘要
ABSTRACT In the field of e‐tourism, online travel agents (OTAs) provide filtering function in platform for booking online tourism products (OTPs); prospective travellers still struggle to articulate the holistic preference from a customized perspective. A group of reviewers' preferences are averaged to represent the performance of OTPs. It offers valuable references but can often conflict with an individual's own preferences. To address these challenges, we develop a hybrid analytical methodology based on online reviews and reach a customized and robust OTPs recommendation. By adopting latent Dirichlet allocation and sentiment analysis, a tourism‐specific hierarchical indicator system is built for selection. Through simulating group preferences by stochastic acceptability analysis, the group judgement policy over any alternative OTPs pair is elicited credibly. To solve the disagreement judgement policies among individuals and group, a group‐individual preference analytics consisting of a heuristic algorithm is proposed to probe the maximal judgement polices with the optimal credibility. Founded on this, a robust aggregation model based on maximax and maximin principles generates a tailor‐made recommendation. A case study on Tripadvisor hotels in Hong Kong demonstrates the feasibility and superiority of the proposed approach in providing robust, preference‐aligned OTP recommendations.
科研通智能强力驱动
Strongly Powered by AbleSci AI