Presenting a method to reduce the sensitivity of incremental clustering algorithms of XML documents based on collective intelligence algorithms

Until now, various methods have been presented for storing and retrieving information of semi-structured documents, most of them are placed in two groups with batch and incremental approach. In the batch or cluster approach, it is assumed that all the documents can be accessed and clustered, and the documents can be processed several times, which increases the execution time of such algorithms. In the incremental approach, all the documents do not exist in one place, but over time, they are provided to the classification method, and from this point of view, the execution time of such algorithms is less compared to the batch method, and as a result, their execution speed is faster. In this research, our proposed method was compared with XCLS and XCLS+ methods in three evaluation criteria: Entropy, Purity and Fscore. The results showed that the proposed method is preferable to the XCLS and XCLS+ methods in terms of Entropy, Purity and Fscore, and it is slightly less efficient than the XCLS+ method only in the Fscore criterion.