基于大数据的旅游目的地情感评价方法探究
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
收稿日期: 2017-01-04
要求修回日期: 2017-04-06
网络出版日期: 2017-06-30
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
Received date: 2017-01-04
Request revised date: 2017-04-06
Online published: 2017-06-30
Copyright
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。
刘逸 , 保继刚 , 朱毅玲 . 基于大数据的旅游目的地情感评价方法探究[J]. 地理研究, 2017 , 36(6) : 1091 -1105 . DOI: 10.11821/dlyj201706008
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
Fig. 1 Literature of tourist emotional analysis based on big-data approach图1 基于大数据的游客情感分析研究进展 |
Tab. 1 The data collection of eight tourist destinations表1 8个旅游目的地的数据采集情况 |
| 监测点 | 评论总数 (条) | 2013年评论数 (条) | 2014年评论数 (条) | 2015年1-9月评论数 (条) | 正面词汇数 (个) | 负面词汇数 (个) |
|---|---|---|---|---|---|---|
| 成都 | 48816 | 4260 | 22563 | 20921 | 173861 | 21149 |
| 黄山 | 13777 | 745 | 6319 | 6411 | 50678 | 8039 |
| 焦作 | 5278 | 455 | 2494 | 2218 | 19309 | 2758 |
| 喀纳斯 | 1836 | 288 | 752 | 702 | 6981 | 949 |
| 洛阳 | 13135 | 1070 | 6283 | 5421 | 48157 | 6385 |
| 西双版纳 | 7310 | 568 | 3768 | 2783 | 29702 | 4095 |
| 阳朔 | 12568 | 1119 | 5098 | 5811 | 47211 | 7700 |
| 张家界 | 18011 | 1134 | 7611 | 8894 | 79664 | 12597 |
| 合计 | 120731 | 9639 | 54888 | 53161 | 455563 | 63672 |
Tab. 2 Emotional evaluation rules of transaction sentence表2 转折句式情感评判规则[45] |
| 句式结构 | 情感极性 |
|---|---|
| 第一类转折词+正面情感词 | 消极 |
| 第一类转折词+奇数重否定+正面情感词 | 积极 |
| 第一类转折词+偶数重否定+正面情感词 | 消极 |
| 第一类转折词+负面情感词 | 积极 |
| 第一类转折词+奇数重否定+负面情感词 | 消极 |
| 第一类转折词+偶数重否定+负面情感词 | 积极 |
| 第二类转折词+正面情感词 | 积极 |
| 第二类转折词+奇数重否定+正面情感词 | 消极 |
| 第二类转折词+偶数重否定+正面情感词 | 积极 |
| 第二类转折词+负面情感词 | 消极 |
| 第二类转折词+奇数重否定+负面情感词 | 积极 |
| 第二类转折词+偶数重否定+负面情感词 | 消极 |
| 第一类转折词+……,第二类转折词+…… | 与只有第一类转折词时相同 |
Tab. 3 Result of emotional evaluation among eight tourist destinations (%)表3 8个旅游目的地的整体情感评价结果(%) |
| 规则 | 评论 | 成都 | 黄山 | 焦作 | 喀纳斯 | 洛阳 | 西双版纳 | 阳朔 | 张家界 |
|---|---|---|---|---|---|---|---|---|---|
| 评分法 | 正面评论 | 82.6 | 89.1 | 84.8 | 80.0 | 85.4 | 79.7 | 71.3 | 81.6 |
| 中性评论 | 14.5 | 9.6 | 12.1 | 18.5 | 12.3 | 15.4 | 20.8 | 14.2 | |
| 负面评论 | 2.9 | 1.3 | 3.1 | 1.5 | 2.3 | 4.9 | 7.9 | 4.2 | |
| 规则A | 正面评论 | 93.9 | 92.3 | 91.7 | 95.2 | 93.9 | 93.8 | 91.1 | 93.8 |
| 中性评论 | 4.4 | 5.7 | 6.3 | 3.8 | 4.4 | 4.3 | 6.1 | 4.3 | |
| 负面评论 | 1.7 | 2.0 | 2.0 | 1.0 | 1.7 | 1.9 | 2.8 | 1.9 | |
| 规则B | 正面评论 | 75.7 | 73.5 | 75.0 | 76.6 | 76.6 | 78.2 | 72.4 | 78.9 |
| 中性评论 | 14.5 | 15.4 | 13.2 | 15.2 | 14.0 | 11.5 | 13.8 | 11.7 | |
| 负面评论 | 9.8 | 11.1 | 11.8 | 8.2 | 9.4 | 10.3 | 13.8 | 9.4 | |
| 规则C1 | 正面评论 | 78.7 | 75.4 | 76.9 | 79.5 | 78.9 | 77.6 | 72.3 | 76.2 |
| 中性评论 | 7.9 | 9.1 | 7.9 | 9.3 | 8.4 | 8.9 | 9.7 | 10.7 | |
| 负面评论 | 13.4 | 15.5 | 15.2 | 11.2 | 12.7 | 13.5 | 18.0 | 13.1 | |
| 规则C2 | 正面评论 | 75.9 | 71.6 | 73.8 | 75.7 | 75.8 | 74.2 | 68.9 | 71.9 |
| 中性评论 | 4.6 | 6.2 | 4.9 | 5.6 | 5.2 | 5.8 | 5.6 | 7.0 | |
| 负面评论 | 19.5 | 22.2 | 21.3 | 18.7 | 19.0 | 19.9 | 25.5 | 20.9 | |
| 规则C3 | 正面评论 | 73.7 | 68.2 | 71.3 | 73.5 | 73.2 | 71.5 | 66.3 | 68.7 |
| 中性评论 | 3.3 | 4.7 | 3.7 | 3.5 | 3.9 | 4.1 | 4.0 | 5.0 | |
| 负面评论 | 23.1 | 27.1 | 25.1 | 22.9 | 22.9 | 24.4 | 29.7 | 26.3 |
Tab. 4 Single-year verification of the monitoring results in 2015表4 2015年监测结果单年度校验表 |
| 旅游目的地 | 监测结果 满意度 | 评分法 正面评论比例 | 规则A 正面评论比例 | 规则B 正面评论比例 | 规则C1 正面评论比例 | 规则C2 正面评论比例 | 规则C3 正面评论比例 |
|---|---|---|---|---|---|---|---|
| 成都 | 0.7950 | 0.8576 | 0.9389 | 0.7848 | 0.7818 | 0.7497 | 0.7243 |
| 黄山 | 0.8270 | 0.9066 | 0.934 | 0.7757 | 0.7548 | 0.7094 | 0.6668 |
| 焦作 | 0.7230 | 0.8877 | 0.9274 | 0.7926 | 0.7615 | 0.7281 | 0.6984 |
| 喀纳斯 | 0.6989 | 0.7536 | 0.9487 | 0.7507 | 0.7635 | 0.7236 | 0.6952 |
| 洛阳 | 0.8130 | 0.8685 | 0.9347 | 0.7805 | 0.7683 | 0.7340 | 0.7080 |
| 西双版纳 | 0.7560 | 0.8286 | 0.9378 | 0.8020 | 0.7614 | 0.7190 | 0.6838 |
| 阳朔 | 0.5994 | 0.7589 | 0.9066 | 0.7472 | 0.6995 | 0.6612 | 0.6307 |
| 张家界 | 0.7080 | 0.8421 | 0.9318 | 0.7993 | 0.7411 | 0.6938 | 0.6613 |
| 绝对方差 | 0.0115 | 0.0411 | 0.0055 | 0.0030 | 0.0035 | 0.0063 |
Tab. 5 The multi-year verification of the monitoring results表5 监测结果多年度校验分析表 |
| 旅游目的地 | 评分法 绝对方差 | 规则A 绝对方差 | 规则B 绝对方差 | 规则C1 绝对方差 | 规则C2 绝对方差 | 规则C3 绝对方差 |
|---|---|---|---|---|---|---|
| 成都 | 0.003397 | 0.030959 | 0.002492 | 0.001927 | 0.001196 | 0.001740 |
| 黄山 | 0.006980 | 0.014129 | 0.005629 | 0.002032 | 0.007080 | 0.013073 |
| 喀纳斯 | 0.016395 | 0.053762 | 0.014592 | 0.013774 | 0.010949 | 0.009868 |
| 阳朔 | 0.013679 | 0.058858 | 0.007633 | 0.005570 | 0.002167 | 0.001050 |
| 张家界 | 0.019668 | 0.065636 | 0.009530 | 0.008771 | 0.003914 | 0.002089 |
| 绝对方差之和 | 0.060120 | 0.223344 | 0.039876 | 0.032074 | 0.025306 | 0.027820 |
The authors have declared that no competing interests exist.
| [1] |
|
| [2] |
[
|
| [3] |
[
|
| [4] |
|
| [5] |
|
| [6] |
[
|
| [7] |
[
|
| [8] |
[
|
| [9] |
[
|
| [10] |
[
|
| [11] |
[
|
| [12] |
[
|
| [13] |
[
|
| [14] |
|
| [15] |
[
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
[
|
| [21] |
|
| [22] |
[
|
| [23] |
[
|
| [24] |
[
|
| [25] |
[
|
| [26] |
|
| [27] |
|
| [28] |
[
|
| [29] |
[
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
|
| [34] |
|
| [35] |
|
| [36] |
[
|
| [37] |
|
| [38] |
[
|
| [39] |
|
| [40] |
|
| [41] |
|
| [42] |
[
|
| [43] |
[
|
| [44] |
[
|
| [45] |
[
|
| [46] |
[
|
| [47] |
[
|
| [48] |
[
|
| [49] |
[
|
| [50] |
[
|
/
| 〈 |
|
〉 |