地理研究 ›› 2017, Vol. 36 ›› Issue (6): 1091-1105.doi: 10.11821/dlyj201706008

所属专题: 地理大数据

• • 上一篇    下一篇

基于大数据的旅游目的地情感评价方法探究

刘逸(), 保继刚(), 朱毅玲   

  1. 中山大学旅游学院,广州 510275
  • 收稿日期:2017-01-04 修回日期:2017-04-06 出版日期:2017-06-30 发布日期:2017-06-30
  • 作者简介:

    作者简介:刘逸(1980- ),男,广东汕头人,讲师,研究方向为旅游价值链与大数据。 E-mail: liuyi89@mail.sysu.edu.cn

  • 基金资助:
    教育部人文社会科学研究青年基金项目(14YJC790083);国家自然科学基金项目(41571137)

Exploring emotion methods of tourism destination evaluation: A big-data approach

Yi LIU(), Jigang BAO(), Yiling ZHU   

  1. School of Tourism Management, Sun Yat-sen University, Guangzhou 510275, China
  • Received:2017-01-04 Revised:2017-04-06 Online:2017-06-30 Published:2017-06-30

摘要:

基于情绪分类取向,通过界定三个旅游文本情感分析的过滤参数:旅游专属词库、语义逻辑规则和情感乘数,构建基于网络大数据的旅游目的地情感评价模型。基于该模型,抓取了120731条游客评论对8个旅游目的地进行评价,并以联合国世界旅游组织旅游可持续发展监测数据作为标准数据进行校验。研究证实三个过滤参数具有一定的科学性,能够较为准确地捕捉到游客对目的地评价的总体情感意象;经过单年度和多年度校验,六类规则的准确度依次为:C2>C1>C3>B>评分法>A,即规则C2下的评价结果与监测结果最为吻合。结论证实了旅游大数据的可用性,为后续的理论推进和实践应用提供了科学依据。

关键词: 大数据, 旅游目的地, 情感评价模型, 情感意象

Abstract:

The contemporary studies of tourism big data are not sufficient to utilize online tourist-generated contents for evaluating tourism destinations, while the content-analysis studies in linguistic studies have yet to have qualified technics for conducting tourism research. In order to bridge this gap between these, this paper constructs an emotion model for evaluating tourism destinations based on tourists' online reviews. This model is composed of three emotional filtering factors including tourism lexicon, grammatical logics and emotional multipliers. The tourism lexicon contains 3507 positive emotional words and 3365 negative emotional words. It is used to depict the general emotional image of a tourist online review by calculating positive and negative words within the review. Every emotional word will be counted as one score, either positive or negative. Grammatical logics contain 13 rules which adjust the positive or negative scores and give the final emotional score of the review. Emotional multiplier in this study is set from three to five. It is used to correct the deviation of exaggerated positive emotions due to the existing pro-positive preference in human emotional expression. This paper collects 120731 pieces of tourists' online reviews among eight tourist destinations (Yangshuo, Zhangjiajie, Huangshan, Chengdu, Luoyang, Kanas, Jiaozuo and Xishuangbanna) and uses this model to evaluate the overall emotional images of these destinations. The result is compared to the questionnaire-based survey data conducted by UNWTO (United Nation, World Tourism Organization) among these destinations from 2013 to 2015. The verification proves that the three emotional filtering factors are effective in mapping emotional images of tourists' online reviews. Based on both single and multiple-year verification, the accuracy of the proposed six sub-models is ranged from high to low as follows: C2>C1>C3>B>Direct Scores>A. This outcome means that the model-based result is the closest to the UNWTO result under the C2 that emotional multiplier is set at 4 and the tourism lexicon and grammatical logics are applied. This paper contributes to the literature by paving alternative ways of destination evaluation and proves the usefulness of tourism big data in geographical studies. This effort will underpin subsequent theoretical and empirical studies in tourism geography.

Key words: big data, tourism destination, emotional evaluation model, emotional image