面向兴趣点推荐的时空序列模式挖掘方法

摘要

本文将通过用户对兴趣点访问所产生的系列位置数据。依据时空数据的特性对原始数据集进行预处理，利用模式挖掘的方法分析兴趣点访问序列中的频繁模式，根据用户当前所在的位置或最近访问序列通过序列分析的方法进行模式匹配，并按照匹配程度给出兴趣点推荐列表。实验结果对比表明数据在预处理后不仅数据量上明显的减少，而且缩短了算法的执行时间，实验的最终结果显示了时空序列模式挖掘与模式匹配方法相结合在兴趣点推荐中的应用是可行的，这为复杂的时空数据分析提供了一种新的视角。%This paper will use users visited points of interest produced a series of location data, According to the characteristics of spatio-temporal data preprocessed of primitive data sets, Using pattern mining methods to analyze Points of interest's sequence of frequent pattern, According to the user's current location or recently visited sequence by sequence analysis method to pattern match- ing, and in accordance with matching degree given points of interest recommended list. The results show that the data comparing pretreatment not only significantly reduced the amount of data on, but also shortened the algorithm of the execution time, The final experimental results show that the spatial and temporal sequential pattern mining and pattern matching methods combining using in points of interest recommended is feasible, this method provides a new perspective for complex spatio-temporal data analysis.