TY - GEN
T1 - Hierarchical multi-armed bandits for discovering hidden populations
AU - Kumar, Suhansanu
AU - Gao, Heting
AU - Wang, Changyu
AU - Chang, Kevin Chen Chuan
AU - Sundaram, Hari
N1 - Funding Information:
VIII. ACKNOWLEDGMENT This material is based upon work supported by the National Science Foundation IIS 16-19302 and IIS 16-33755, Zhejing University ZJU Research 083650, Futurewei Technologies HF2017060011 and 094013, UIUC OVCR CCIL Planning Grant 434S34, UIUC CSBS Small Grant 434C8U, and Advanced Digital Sciences Center Faculty Grant. REFERENCES
Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019/8/27
Y1 - 2019/8/27
N2 - This paper proposes a novel algorithm to discover hidden individuals in a social network. The problem is increasingly important for social scientists as the populations (e.g., individuals with mental illness) that they study converse online. Since these populations do not use the category (e.g., mental illness) to self-describe, directly querying with text is non-trivial. To by-pass the limitations of network and query re-writing frameworks, we focus on identifying hidden populations through attributed search. We propose a hierarchical Multi-Arm Bandit (DT-TMP) sampler that uses a decision tree coupled with reinforcement learning to query the combinatorial attributed search space by exploring and expanding along high yielding decision-tree branches. A comprehensive set of experiments over a suite of twelve sampling tasks on three online web platforms, and three offline entity datasets reveals that DT-TMP outperforms all baseline samplers by upto a margin of 54% on Twitter and 48% on RateMDs. An extensive ablation study confirms DT-TMP’s superior performance under different sampling scenarios.
AB - This paper proposes a novel algorithm to discover hidden individuals in a social network. The problem is increasingly important for social scientists as the populations (e.g., individuals with mental illness) that they study converse online. Since these populations do not use the category (e.g., mental illness) to self-describe, directly querying with text is non-trivial. To by-pass the limitations of network and query re-writing frameworks, we focus on identifying hidden populations through attributed search. We propose a hierarchical Multi-Arm Bandit (DT-TMP) sampler that uses a decision tree coupled with reinforcement learning to query the combinatorial attributed search space by exploring and expanding along high yielding decision-tree branches. A comprehensive set of experiments over a suite of twelve sampling tasks on three online web platforms, and three offline entity datasets reveals that DT-TMP outperforms all baseline samplers by upto a margin of 54% on Twitter and 48% on RateMDs. An extensive ablation study confirms DT-TMP’s superior performance under different sampling scenarios.
UR - http://www.scopus.com/inward/record.url?scp=85078840528&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85078840528&partnerID=8YFLogxK
U2 - 10.1145/3341161.3342880
DO - 10.1145/3341161.3342880
M3 - Conference contribution
AN - SCOPUS:85078840528
T3 - Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2019
SP - 145
EP - 153
BT - Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2019
A2 - Spezzano, Francesca
A2 - Chen, Wei
A2 - Xiao, Xiaokui
PB - Association for Computing Machinery
T2 - 11th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2019
Y2 - 27 August 2019 through 30 August 2019
ER -