【摘要】 为解决频繁轨迹模式挖掘中单一最小支持度带来的问题,提出一种多最小支持度的频繁序列挖掘算法,根据获取的用户历史轨迹数据确定用户多最小支持度获取模型。由于仅通过Prefix Span算法挖掘出用户的历史频繁轨迹模式,无法了解用户在一段时间内的地点偏好变化,通过动态加权的方式结合之前挖掘出的用户频繁轨迹模式得到用户在不同时期的地点偏好变化,利用序列压缩和序列匹配减少用户频繁轨迹模式的存储空间。通过实例挖掘,验证了改进算法的有效性。更多还原

【Abstract】 To solve the problem caused by single minimum support in frequent track pattern mining,a frequent sequence mining algorithm with multiple minimum support was proposed, which determined the users’ multiple minimum support acquisition model based on the acquired users’ historical track data.It is unable to understand the users’ location preferences change over a period of time only through the PrefixSpan algorithm for mining the frequent users’ history trajectory model.Therefore,the location preference changes of users in different periods were obtained by combining the previously mined frequent trajectory patterns in a dynamic weighted way.Sequence compression and sequence matching were used to greatly reduce the storage space of users’ frequent trace patterns.The effectiveness of the improved algorithm was verified by example mining.更多还原