| The advancements of the wireless communication technique and popularity of mobile devices, have contributed to a wide of applications of Location-Based Service(LSB) ,and the researches on Location-Based Service(LSB) have been emerging in recent years. It is the mining and prediction of mobile movements and associated transactions that is one of active topics. Most studies have focused on the behavior of the log to mine the mobile patterns. However, most studies do not consider the temporal period to have a effect on mobile behaviors among user.Along with the advancements of data mining theory and the technology, it become more effectively find the useful information from the large of data. R.Agrawal proposed pattern mining in 1995. After that, a lot of scholars have proposed many of algorithms to improve the efficiency of sequence mining. This paper analyzes the main drawback of Apriori based on time algorithm. The Apriori-based algorithm may generate a really large set of candidate sequences. A long candidate sequences grow from a combination of short ones, and the number of such candidate sequences is exponential to the length of the sequential patterns. The Apriori based on time algorithm have to multiple scans of databases in order to calculation the support, this sequence can be called sequential patterns, only when the support is greater than the set minimum support. The sequence patterns mean high of regularity of the movement. This paper apply the prefixSpan algorithm to improve the Apriori based on time algorithm, the prefixSpan do not like the Apriori algorithm to generate the large set of candidate sequences, therefore, the search space is smaller. The major cost of prefixSpan is the construction of projected databases, but compared to the original projection database, the projected databases is smaller, because the sequences in the projected databases is postfix with regard to prefix, the number of sequences in a projected database will become quite small when prefix grows. |
PDF Full Text Request