基于单点地名匹配和局部地名筛选的推特用户定位方法

doi:10.11959/j.issn.2096-109x.2023053

网络与信息安全学报 ›› 2023, Vol. 9 ›› Issue (4): 53-63.doi: 10.11959/j.issn.2096-109x.2023053

• 学术论文 • 上一篇

基于单点地名匹配和局部地名筛选的推特用户定位方法

薛锦¹^,², 袁福祥², 刘毅敏², 张萌², 乔亚琼²^,³, 罗向阳²

¹ 郑州大学网络空间安全学院，河南郑州 450003
² 河南省网络空间态势感知重点实验室，河南郑州 450001
³ 华北水利水电大学信息工程学院，河南郑州 450045

修回日期:2023-04-18 出版日期:2023-08-01 发布日期:2023-08-01
作者简介:薛锦（1996- ），男，内蒙古乌兰察布人，郑州大学硕士生，主要研究方向为社交网络定位与数据挖掘
袁福祥（1991- ），男，山东济宁人，博士，信息工程大学讲师，主要研究方向为网络目标定位、网络拓扑分析与网络空间测绘
刘毅敏（1995- ），女，山东烟台人，信息工程大学博士生，主要研究方向为网络安全与社交网络数据分析
张萌（1996- ），女，河南偃师人，信息工程大学博士生，主要研究方向为数据挖掘和社交网络分析
乔亚琼（1981- ），女，河南开封人，博士，华北水利水电大学讲师，主要研究方向为数据挖掘与社交网络分析
罗向阳（1978- ），男，湖北荆门人，博士，信息工程大学教授、博士生导师，主要研究方向为网络与信息安全
基金资助:
国家自然科学基金(U1804263);国家自然科学基金(U2172435);国家自然科学基金(62272163);国家重点研发计划(2022YFB3102900);中原科技创新领军人才项目(214200510019);河南省科技攻关项目(222102210036);河南省自然科学青年基金(222300420230)

Twitter user geolocation method based on single-point toponym matching and local toponym filtering

Jin XUE¹^,², Fuxiang YUAN², Yimin LIU², Meng ZHANG², Yaqiong QIAO²^,³, Xiangyang LUO²

¹ School of Cyber Science and Engineering, Zhengzhou University, Zhengzhou 450003, China
² Henan Key Laboratory of Cyberspace Situation Awareness, Zhengzhou 450001, China
³ School of Information Engineering, North China University of Water Resources and Electric Power, Zhengzhou 450045, China

Revised:2023-04-18 Online:2023-08-01 Published:2023-08-01
Supported by:
The National Natural Science Foundation of China(U1804263);The National Natural Science Foundation of China(U2172435);The National Natural Science Foundation of China(62272163);The National Key Research and Development Program of China(2022YFB3102900);Zhongyuan Science and Technology Innovation Leading Talent Project of China(214200510019);The Key Science and Technology Project of Henan Province(222102210036);The Henan Province Science Foundation for Youths(222300420230)

摘要/Abstract

摘要：

用户推文中的地名信息是定位推特用户的重要基础数据之一，但现有推特用户定位方法提取的地名在数量和可靠性方面均存在欠缺，影响了用户定位准确性。提出基于单点地名匹配和局部地名筛选的推特用户定位方法。设计一种基于地名位置聚集度的地名类型判别算法，根据位置分布的聚集程度构建单点地名库，获取推文中更多可靠地名；提出一种基于用户位置聚集度的局部地名筛选算法，分别以地名经纬度和用户平均经纬度为中心，计算用户位置聚集度，筛选更高聚集度、更可靠的局部地名；基于用户社交关系、用户对地名的提及关系构建用户-地名异构图，并利用图表示学习和神经网络定位用户。基于常用公开数据集GEOTEXT和TW-US进行大量用户定位实验，并与HGNN、ReLP、GCN等9种现有推特用户位置推断典型方法进行了对比，结果表明，所提方法对推特用户的位置推断准确率具有明显优势，相比9种现有典型方法，在GEOTEXT数据集上，平均误差降低了7.3～342.8 km，中位数误差降低了2.4～354.4 km，大地区级定位准确率提高了1.3%～26.3%；在TW-US数据集上，平均误差降低了8.6～246.6 km，中位数误差降低了5.7～149.7 km，大地区级定位准确率提高了1.5%～20.5%。

关键词: 用户定位, 用户生成内容, 地名, 社交媒体

Abstract:

The availability of accurate toponyms in user tweets is crucial for geolocating Twitter users.However, existing methods for locating Twitter users often suffer from limited quantity and reliability of acquired toponyms, thus impacting the accuracy of user geolocation.To address this issue, a twitter user geolocation method based on single-point toponym matching and local toponym filtering was proposed.A toponym type discriminating algorithm based on the aggregation degree of locations of the toponym was designed.In the proposed algorithm, a single-point toponym database was generated to provide more reliable toponyms extracted from tweets.Then, according to a proposed local place name filtering algorithm based on the aggregation degree of user location, the aggregation degree of user location centered on the longitude and latitude of toponyms and the average longitude and latitude of users were calculated.This process helped in extracting local toponyms with a high aggregation degree, which enhances the reliability of toponyms used in geolocation.Finally, a user-toponym heterogeneous graph was constructed based on user social relationships and user mentions of toponyms, and users were located by graph representation learning and neural networks.A large number of user geolocation experiments were conducted based on two commonly used public datasets in this field, namely GEOTEXT and TW-US.Comparisons with nine existing typical methods for Twitter user geolocation, including HGNN, ReLP, and GCN, demonstrate that our proposed method achieves significantly higher geolocation accuracy.On the GEOTEXT dataset, the average error is reduced by 7.3～342.8 km, the median error is reduced by 2.4～354.4 km, and the accuracy of large area-level geolocation is improved by 1.3%～26.3%.On the TW-US dataset, the average error is reduced by 8.6～246.6 km, the median error is reduced by 5.7～149.7 km, and the accuracy of large area-level geolocation is improved by 1.5%～20.5%.

Key words: user geolocation, user-generated text, toponym, social media

中图分类号:

TP391

薛锦, 袁福祥, 刘毅敏, 张萌, 乔亚琼, 罗向阳. 基于单点地名匹配和局部地名筛选的推特用户定位方法[J]. 网络与信息安全学报, 2023, 9(4): 53-63.

Jin XUE, Fuxiang YUAN, Yimin LIU, Meng ZHANG, Yaqiong QIAO, Xiangyang LUO. Twitter user geolocation method based on single-point toponym matching and local toponym filtering[J]. Chinese Journal of Network and Information Security, 2023, 9(4): 53-63.

图/表 10

表1

图1

图2

表2

表3

表4

表5

表6

推特用户定位性能对比Table 6 Comparison of twitter user geolocation performance"

模型		GEOTEXT数据集			TW-US数据集
模型		大地区级定位准确率	平均误差/km	中位数误差/km	大地区级定位准确率	平均误差/km	中位数误差/km
基于用户	HierLR^[10]	—	—	—	48%	656	191
文本方法	MLP4Geo^[11]	38%	844	389	54%	554	120
基于社交	MADCEL-W^[21]	58%	586	60	54%	705	116
关系方法	GCN-LP^[16]	58%	576	56	53%	563	126
	MENET^[17]	59.1%	570	58	50.5%	474	157
	GCN^[16]	60%	546	45	65%	485	71
多源数据	ReLP^[20]	62%	527	37	67%	418	47
融合方法	MetaGeo^[23]	62%	533	42	63%	479	70
	HGNN^[24]	63%	508.5	37	65%	423.2	59.5
	GEOTOP	$64 . 3 %$	$501 . 2$	$34 . 6$	$68 . 5 %$	$409 . 4$	$41 . 3$

表6

表7

表8

参考文献 5

[5]	GU Q Y , JU C H , WU G X . Social network link prediction method based on subgraph evolution and improved ant colony optimization algorithm[J]. Journal on Communications, 2020,41(12): 21-35.
[6]	CHENG Z Y , CAVERLEE J , LEE K . A content-driven framework for geolocating microblog users[J]. ACM Transactions on Intelligent Systems and Technology, 2013,4(1): 1-27.
[7]	EISENSTEIN J , CONNOR B O , SMITH N A ,et al. A latent variable model for geographic lexical variation[C]// Proceedings of the 2010 Conference on Empirical Methods in Natural Language Pro-cessing (EMNLP ’10). 2010: 1277-1287.
[8]	HAN B , COOK P , BALDWIN T . Geolocation prediction in social media data by finding location indicative words[C]// Proceedings of the 24th International Conference on Computational Linguistics(COLING). 2012: 1045-1062.
[9]	JONES K S . A statistical interpretation of term specificity and its application in retrieval[J]. Journal of Documentation, 1988,60(5): 493-502.
[10]	WING B , BALDRIDGE J . Hierarchical discriminative classification for text based geolocation[C]// Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing(EMNLP). 2014: 336-348.
[11]	RAHIMI A , COHN T , BALDWIN T . A neural model for user geolocation and lexical dialectology[C]// Proceedings of Annual Meeting of the Association for Computational Linguistics (ACL). 2017: 209-216.
[12]	DAVIS J R C , PAPPA G L , OLIVEIRA D R R ,et al. Inferring the location of twitter messages based on user relationships[J]. Transactions in GIS, 2011,15(6): 1361-1682.
[13]	KONG L , LIU Z , HUANG Y . Spot:Locating social media users based on social network context[J]. Proceedings of the VLDB Endowment, 2014,7(13): 1681-1684.
[14]	JURGENS D . That's what friends are for:Inferring location in online social media platforms based on social relationships[C]// Proceedings of the Seventh International Conference on Weblogs and Social Media. 2013: 273-282.
[15]	MC-GEE J , CAVERLEE J , CHENG Z Y . Location prediction in social media based on tie strength[C]// Proceedings of the 22nd ACM International Conference on Information ＆ Knowledge Management (CIKM ’13). 2013: 459-468.
[16]	RAHIMI A , COHN T , BALDWIN T . Semi-supervised user geolocation via graph convolutional networks[C]// Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics. 2018: 2009-2019.
[17]	DO T H , NGUYEN D M , TSILIGIANNI E ,et al. Twitter user geolocation using deep multiview learning[C]// Proceeding of the 2018 IEEE International Conference on Acoustics,Speech and Signal Processing (ICASSP). 2018: 6304-6308.
[18]	LE Q , MIKOLOV T . Distributed representations of sentences and documents[C]// Proceedings of the 31st International Conference on Machine Learning. 2014: 1188-1196.
[19]	GROVER A , LESKOVEC J . node2vec:scalable feature learning for networks[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2016: 855-864.
[20]	TIAN H C , ZHANG M , LUO X Y ,et al. Twitter user location inference based on representation learning and label propagation[C]// Proceedings of The Web Conference. 2020: 2648-2654.
[21]	RAHIMI A , COHN T , BALDWIN T . Twitter user geolocation using a unified text and network prediction model[C]// Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics-7th International Joint Conference on Natural Language Processing(NAACL-HLT ’15). 2015: 630-636.
[22]	王凯, 余伟, 吴敏 ,等. 一种大数据环境下的在线社交媒体位置推断方法[J]. 软件学报, 2015,26(11): 2951-2963.
	WANG K , YU W , WU M ,et al. Location inference method in online social media with big data[J]. Journal of Software, 2015,26(11): 2951-2963.
[23]	ZHOU F , QI X , ZHANG K ,et al. MetaGeo:a general framework for social user geolocation identification with few-shot learning[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022, 1-15.
[24]	ZHOU F , WANG T , ZHONG T ,et al. Identifying user geolocation with hierarchical graph neural networks and explainable fusion[J]. Information Fusion, 2022,81: 1-13.
[25]	ROLLER S , SPERIOSU M , RALLAPALLI S ,et al. Supervised text based geolocation using language models on an adaptive grid[C]// Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. 2012: 1500-1510.
[1]	LIANG L , MALBASA V , VUCETIC S . Spatial scan for disease mapping on a mobile population[C]// AAAI Conference on Artificial Intelligence. 2014: 431-437.
[2]	TUMASJAN A , SPRENGER T O , SANDNER P G ,et al. Predicting elections with Twitter:what 140 characters reveal about political sentiment[C]// International Conference on Weblogs and Social Media. 2010: 178-185.
[3]	HALIMI A , AYDAY E . Profile matching across online social networks[C]// 22nd International Conference on Information and Communications Security. 2020: 54-70.
[4]	顾秋阳, 吴宝, 池仁勇 . 基于高阶路径相似度的复杂网络链路预测方法[J]. 通信学报, 2021,42(7): 61-69.
	GU Q Y , WU B , CHI R Y . Link prediction method based on the similarity of high path[J]. Journal on Communications, 2021,42(7): 61-69.
[5]	顾秋阳, 琚春华, 吴功兴 . 基于子图演化与改进蚁群优化算法的社交网络链路预测方法[J]. 通信学报, 2020,41(12): 21-35.

符号	含义
TP	地名集合
TP ^I	间接地名集合
TP ^D	直接地名集合
LIW	位置指示词集合
k	基于k维树划分后每个位置标签包含的用户数
V^L	有标签用户集合
V^U	目标用户集合
G'	用户-地名图
X	用户推文集合
P^V	用户位置的集合
P^U	目标用户位置的集合
P^L	有标签用户位置的集合
P^TP	地名位置的集合
P	用户和地名的位置集合
f₁	提取直接地名的阈值
f₂	提取间接地名的阈值
s	全局名人过滤阈值

项目	覆盖州数/个	覆盖城市数/个	地名数/个
细粒度地名	50	307	3 035 437
单点地名	50	311	2 043 527

参数	GEOTEXT	TW-US
k	50人	2 400人
输入层维度	128	128
隐藏层维度	256	256
输出层维度	129	256
λ₁	0.5	0.5
λ₂	0.5	0.5
β₁	0.5	0.5
β₂	0.5	0.5
d	150 km	150 km
f₁	0.7	0.75
f₂	0.8	0.82
s	5人	15人
p	4	4
q	0.25	0.25

数据集	训练集用户数/个	验证集用户数/个	测试集用户数/个
GEOTEXT^[7]	5 685	1 895	1 895
TW-US^[25]	1 366 766	10 000	10 000

数据集	用户数/个	推文数/条	用户提及数/个
GEOTEXT^[7]	9 475	377 504	232 598
TW-US^[25]	449 200	38 036 187	8 128 105

基于单点地名匹配和局部地名筛选的推特用户定位方法

Twitter user geolocation method based on single-point toponym matching and local toponym filtering

在线阅读

pdf下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 5

相关文章 3

Metrics

推荐阅读 0

地名	指示位置	UL/TL	大地区级定位准确率
paerdegat	40.629, -73.906	1.00	100%
florin mall	38.497, -21.433	1.00	100%
Florida Agricultural and Mechanical University	30.426, -84.285	1.00	100%
suitland	38.848, -76.923	0.91	100%
anaheim stadium	33.8, -117.882	1.00	100%
kearny mesa	32.921, -117.117	0.87	100%
lansing mall	42.743, -84.625	0.86	100%
wheaton mall	38.803, -77.134	0.93	100%

[1]	黄诗瑀, 叶锋, 黄添强, 李伟, 黄丽清, 罗海峰. 人脸伪造与检测中的对抗攻防综述[J]. 网络与信息安全学报, 2023, 9(4): 1-15.
[2]	郑洪浩, 郝一诺, 于洪涛, 李邵梅, 吴翼腾. 基于改进Transformer的社交媒体谣言检测[J]. 网络与信息安全学报, 2022, 8(4): 168-174.
[3]	时文旗, 罗向阳, 郭家山. 基于加权最小二乘的社交网络用户定位方法[J]. 网络与信息安全学报, 2022, 8(3): 41-52.

模型	大地区级定位准确率	平均误差/km	中位数误差/km	覆盖率
第一个实验	61.7%	518.3	54.2	98.2%
第二个实验	82%	83.7	7.4	3.3%
GEOTOP	64.3%	501.2	34.6	98.4%