Conceptual lattice-based natural noise correction mechanism in knowledge recommendation

In recent years, recommender systems have been widely used. However, the accuracy of recommendations is vulnerable to noise, particularly that which arises naturally from various sources. Therefore, mitigating the effect of natural noise on recommendation precision is a significant issue worthy of investigation. The concept lattice proves to be useful in thoroughly exploring the implicit connections among data, thereby improving the interpretability of the findings. In this paper, we propose a natural noise correction method based on concept lattice detection in the context of knowledge recommendation. The approach initially sorts students and knowledge points into strong, average, and weak categories to identify natural noise. Subsequently, a concept lattice-based approach is utilized to rectify these natural noises. Finally, unrated items are anticipated from the resultant natural noise-free dataset. Experimental results indicate that the method exhibits high recommendation accuracy and performs well even in the case of sparse data.