一种基于情景相似度的多维信息推荐新方法研究

A New Method for Multi-dimensional Information Recommendation Based on the Similarity of Context

传统信息推荐方法只涉及到用户和项目（资源）这两个因素，忽略了情景因素，导致推荐效率比较低，而多维信息推荐在推荐的过程中考虑情景因素对用户行为的影响，动态捕捉用户兴趣在不同情景下的变化，从而大大地提高了信息推荐的效果，向用户提供更加个性化、智能化的推荐结果。本论文首先分析传统信息推荐的主要流程，然后提出了情景以及情景相似度这两个新的概念，构建了基于情景相似度的多维信息推荐系统模型，研制了基于情景相似度的多维信息推荐算法，并通过实验研究的方法验证了论文所提出的新算法的高效性与优越性。

Traditional recommendation methods only involve user and item （resource） , ignoring context, which causes the low efficiency. However, multi-dimensional information recommendation（MDIR） considers the influences of the context on the users, dynamically catching the changes of users＇ interests in different contexts. MDIR can provide more individualized and intellectualized results. This paper firstly analyses the processes of the traditional information recommendation. Secondly the paper brings forth two new concepts： context and similarity of context. Thirdly, based on the similarity of context, the paper builds up a new multi-dimensional information recommendation model and designs the new multi-dimensional recommendation algorithm. Finally, the paper adopts an experiment to test and verify the high efficiency of the new algorithm.