Hidden populations refer to the minority groups that not well-known to the public. Traditional statistical survey methods are difficult to apply in the study of hidden populations because of that the hidden populations individuals are very troublesome to be found and they are not willing to share the inner opinion with the others. On the other hand, with the development of the Web 2.0, the hidden populations gather and share their views in online social networks due to the openness and anonymity of the Internet. So, this paper analyzes the behavioral characteristics of the hidden populations based on their data in online social networks. This paper uses the lesbian population as an example and analyzes the behavioral characteristics of lesbian by analyzing the data of the lesbian population in Douban Group. First, the activity data on lesbian are collected from Douban Group. Second, behavior characteristics of lesbian are analysed, the regional characteristic, temporal characteristic and text characteristic are mined out by big data method. Third, a lesbian recognition model is proposed based on the above analytical characteristics, and the effectiveness of the recognition model is varified by experiment study. The research of this paper is helpful to understand the behavioral characteristics of hidden populations deeply, and provides decision-making basis of management and service for hidden populations.
Li, Minglei; Jiang, Guoyin; Liu, Wenping; and Lei, Junli, "Behavior Analysis and Recognition of Hidden Populations in Online Social Network Based on Big Data Method" (2021). WHICEB 2021 Proceedings. 51.