A Textual Data-Oriented Method for Doctor Selection in Online Health Communities

As doctor–patient interactive platforms, online health communities (OHCs) offer patients massive information including doctor basic information and online patient reviews. However, how to develop a systematic framework for doctor selection in OHCs according to doctor basic information and online patient reviews is a challenged issue, which will be explored in this study. For doctor basic information, we define the quantification method and aggregate them to characterize relative influence of doctors. For online patient reviews, data analysis techniques (i.e., topics extraction and sentiment analysis) are used to mine the core attributes and evaluations. Subsequently, frequency weights and position weights are respectively determined by a frequency-oriented formula and a position score-based formula, which are integrated to obtain the final importance of attributes. Probabilistic linguistic-prospect theory-multiplicative multiobjective optimization by ratio analysis (PL-PT-MULTIMOORA) is proposed to analyze patient satisfactions on doctors. Finally, selection rules are made according to doctor influence and patient satisfactions so as to choose optimal and suboptimal doctors for rational or emotional patients. The designed textual data-driven method is successfully applied to analyze doctors from Haodf.com and some suggestions are given to help patients pick out optimal and suboptimal doctors.