TY - GEN
T1 - Online community conflict decomposition with pseudo spatial permutation
AU - Chen, Yunmo
AU - Ye, Xinyue
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2019.
PY - 2019
Y1 - 2019
N2 - Online communities are composed of individuals sharing similar opinions or behavior in the virtual world. Facilitated by the fast development of social media platforms, the expansion of online communities have raised many attentions among the researchers, business analysts, and decision makers, leading to a growing list of literature studying the interactions especially conflicts in the online communities. A conflict is often initiated by one community which then attacks the other, leading to an adversarial relationship and worse social impacts. Many studies have examined the origins and process of online community conflict while failing to address the possible spatial effects in their models. In this paper, we explore the prediction of online community conflict by decomposing and analyzing its prediction error taking geography into accounts. Grounding on the previous natural language processing based model, we introduce pseudo spatial permutation to test the model expressiveness with geographical factors. Pseudo spatial permutation employs different geographical distributions to sample from and perturbs the model using the pseudo geographical information to examine the relationship between online community conflict and spatial distribution. Our analysis shows that the pseudo spatial permutation is an efficient approach to robustly test the conflict relation learned by the prediction model, and also reveals the necessity to incorporate geographical information into the prediction. In conclusion, this work provides a different aspect of analyzing the community conflict that does not solely rely on the textual communication.
AB - Online communities are composed of individuals sharing similar opinions or behavior in the virtual world. Facilitated by the fast development of social media platforms, the expansion of online communities have raised many attentions among the researchers, business analysts, and decision makers, leading to a growing list of literature studying the interactions especially conflicts in the online communities. A conflict is often initiated by one community which then attacks the other, leading to an adversarial relationship and worse social impacts. Many studies have examined the origins and process of online community conflict while failing to address the possible spatial effects in their models. In this paper, we explore the prediction of online community conflict by decomposing and analyzing its prediction error taking geography into accounts. Grounding on the previous natural language processing based model, we introduce pseudo spatial permutation to test the model expressiveness with geographical factors. Pseudo spatial permutation employs different geographical distributions to sample from and perturbs the model using the pseudo geographical information to examine the relationship between online community conflict and spatial distribution. Our analysis shows that the pseudo spatial permutation is an efficient approach to robustly test the conflict relation learned by the prediction model, and also reveals the necessity to incorporate geographical information into the prediction. In conclusion, this work provides a different aspect of analyzing the community conflict that does not solely rely on the textual communication.
KW - Neural network
KW - Online community
KW - Spatial permutation
KW - Spatial social network
KW - Text mining
UR - http://www.scopus.com/inward/record.url?scp=85077771986&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85077771986&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-34980-6_28
DO - 10.1007/978-3-030-34980-6_28
M3 - Conference contribution
AN - SCOPUS:85077771986
SN - 9783030349790
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 246
EP - 255
BT - Computational Data and Social Networks - 8th International Conference, CSoNet 2019, Proceedings
A2 - Tagarelli, Andrea
A2 - Tong, Hanghang
PB - Springer
T2 - 8th International Conference on Computational Data and Social Networks, CSoNet 2019
Y2 - 18 November 2019 through 20 November 2019
ER -