Automatic Opioid User Detection From Twitter: Transductive Ensemble Built On Different Meta-graph Based Similarities Over Heterogeneous Information Network

Opioid (e.g., heroin and morphine) addiction has become one of the largest and deadliest epidemics in the United States. To combat such deadly epi- demic, in this paper, we propose a novel frame- work named HinOPU to automatically detect opi- oid users from Twitter, which will assist in sharp- ening our understanding toward the behavioral pro- cess of opioid addiction and treatment. In HinOPU, to model the users and the posted tweets as well as their rich relationships, we introduce structured heterogeneous information network (HIN) for rep- resentation. Afterwards, we use meta-graph based approach to characterize the semantic relatedness over users; we then formulate different similari- ties over users based on different meta-graphs on HIN. To reduce the cost of acquiring labeled sam- ples for supervised learning, we propose a trans- ductive classification method to build the base clas- sifiers based on different similarities formulated by different meta-graphs. Then, to further improve the detection accuracy, we construct an ensemble to combine different predictions from different base classifiers for opioid user detection. Comprehen- sive experiments on real sample collections from Twitter are conducted to validate the effectiveness of HinOPU in opioid user detection by compar- isons with other alternate methods.