User's Cognitive-Behavior-Based Preference Access Under Disease-Specific Online Medical Inquiry Text Mining
针对疾病特定在线医疗问询文本，提出一种基于用户认知行为的偏好挖掘方法，通过考虑上下文信息对认知行为的影响，显著提高了偏好预测的精度。
Substantial real cases can be formed from disease-specific online medical inquiry texts. Therefore, user preferences can constitute potentially rich commercial medical value and provide decision support for medical service recommendations. It is necessary to mine user preferences in disease-specific online medical inquiry texts. However, user preferences will change with cognitive behavior decisions in the context of disease-specific inquiry texts. More importantly, the context of disease-specific online inquiry texts in which user preferences are located will affect users' cognitive behavior decisions in real time. However, the existing preference access methods have relatively low precision since they fail to consider the inherent connection between different users’ cognitive behaviors in various contexts and their preferences. We expanded the contextual perception preference model to propose a cognitive-behavior-based method to assess user preference. The contextual information (in online medical inquiry texts on various disease topics) and the impacts of user and text attributes on their cognitive behaviors were abstracted into concept models (including level, usefulness, risk, and effectiveness cognition) to obtain the mutual influences and adjusted relationships. Moreover, more accurate user cognition references were obtained within the multidimensional text space and multidimensional disease space. Based on a large-volume real-world dataset, our experiments revealed the AP@R as the evaluation standard. The method proposed in this article, compared with the contextual perception and ELPCAP algorithms, significantly improves the precision of preference prediction after the addition of cognitive behaviors, suggesting that the method can effectively mine the user cognitive behavior-preference relationships.