Research On Precision Information Recommendation For Online Health Community Based On Mental Accounting

The online health communities provide an effective communication channel for user groups such as patients and doctors.More and more people are seeking solutions to health problems in online health communities.However,the problem of information overload often requires users to spend a lot of time and effort to find the information they need.Recommender systems are a classic way to alleviate information overload.Most recommendation models are built on the common characteristics of users,and recommend content that is most similar to their historical behavior from a rational perspective.However,the traditional rationality assumption has limitations and cannot fully explain the non-substitutability,information avoidance,and risk aversion phenomena in health information behavior.The mental accounting theory reveals the potential psychological structure that individuals use to categorize,encode,and evaluate results when making complex decisions,thereby reducing decision difficulty and explaining the limited rationality phenomenon in decision-making.Therefore,this study introduces the mental accounting theory,considers users’ psychological cognitive processes from a limited rationality perspective,and explicitly models the classification and operation of user participation benefits in online health communities according to mental accounts,constructing a precision information recommendation model that meets users’ psychological needs.This study constructs a precision information recommendation model for online health communities based on mental accounting.The core framework of the model is based on three important parts of mental accounting: the classification and budget control of resources,the evaluation frequency and selection framework of mental accounts,and the gain and loss operation and valuation process of accounts.Based on these three parts,three model components were designed: a topic-based profit and loss account based on LDA topic model,an account scope dynamic delineation based on time decay function,and a profit and loss perception and evaluation based on value function.For online forum data,topic post features and user features were extracted,and a profit and loss calculation module on the threads user participation in was constructed based on motivational theory.Finally,all the above modules were combined to create a precision information recommendation model for online health communities based on mental accounting.To validate the performance of the model,this study collected data from the diabetes forum "Sweet Home" for empirical analysis and used Recall and NDCG indicators to evaluate the performance of the baseline model and the model proposed in this study.The results showed that the performance of the model proposed in this study was better than that of all baseline models.Compared with the best performing model among the baseline models,the Recall score of this study’s model was increased by 19.42%,28.43%,and 30.30% respectively under the recommendation list lengths of K=5,K=10,and K=20,and the NDCG was increased by 2.62%,8.59%,and 12.36% respectively.In the ablation experiment,the performance of the original model was higher than that of the three variant models that eliminated time decay,value function,and individual differences,respectively,which confirmed the effectiveness of the model based on mental accounting modules.This study enriched the theoretical perspective of information recommendation research,constructed an information recommendation model considering users’ psychological cognitive processes based on the mental accounting theory,and verified the effectiveness of introducing psychological factors from the perspective of bounded rationality to improve recommendation models.It has made positive explorations for future recommendation systems based on modeling human cognitive biases.