TY - JOUR
T1 - Tweets can tell
T2 - activity recognition using hybrid gated recurrent neural networks
AU - Cui, Renhao
AU - Agrawal, Gagan
AU - Ramnath, Rajiv
N1 - Publisher Copyright:
© 2020, Springer-Verlag GmbH Austria, part of Springer Nature.
Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020/12/1
Y1 - 2020/12/1
N2 - This paper presents techniques to detect the “offline” activity (such as dining, shopping, or entertainment) a person is engaged in when she is tweeting , in order to create a dynamic profile of the user, for uses such as better targeting of advertisements. To this end, we present a hybrid gated recurrent neural network (GRNN)-based model for rich contextual learning. Specifically, the study and construction of the hybrid model are applied to two types of GRNNs, namely LSTM and GRU networks. In the process, we study the effects of applying and combining multiple contextual modeling methods with different contextual features. Our hybrid model outperforms a set of baselines and state-of-the-art methods. Finally, this paper presents an orthogonal validation using a real-world application. Our model generates offline activity analysis for the followers of several well-known accounts, and the result is quite representative of the expected characteristics of these accounts.
AB - This paper presents techniques to detect the “offline” activity (such as dining, shopping, or entertainment) a person is engaged in when she is tweeting , in order to create a dynamic profile of the user, for uses such as better targeting of advertisements. To this end, we present a hybrid gated recurrent neural network (GRNN)-based model for rich contextual learning. Specifically, the study and construction of the hybrid model are applied to two types of GRNNs, namely LSTM and GRU networks. In the process, we study the effects of applying and combining multiple contextual modeling methods with different contextual features. Our hybrid model outperforms a set of baselines and state-of-the-art methods. Finally, this paper presents an orthogonal validation using a real-world application. Our model generates offline activity analysis for the followers of several well-known accounts, and the result is quite representative of the expected characteristics of these accounts.
KW - Activity recognition
KW - Classification
KW - Neural network
KW - Social network
KW - User profiling
UR - http://www.scopus.com/inward/record.url?scp=85081032207&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85081032207&partnerID=8YFLogxK
U2 - 10.1007/s13278-020-0628-0
DO - 10.1007/s13278-020-0628-0
M3 - Article
AN - SCOPUS:85081032207
SN - 1869-5450
VL - 10
JO - Social Network Analysis and Mining
JF - Social Network Analysis and Mining
IS - 1
M1 - 16
ER -