Pages: 7-15

Abstract

This research aims to improve and summarize the text based on clustering based on collective intelligence algorithm. The algorithm that is calculated in this way is based on the binary particle aggregation algorithm. Each particle size in this algorithm is measured with a fitness function, but instead of using the speed equation, the new particle position is calculated. What has been done in this study is to provide a general hybrid model of the two TD -IDF algorithms along with the PSO multi-factor clustering Which covers the main body. The proposed method is based on the method of weighting the TF-IDF mechanism, Mr. Salton, which uses the repetition of the document's words and queries to calculate the weight. The main idea is to specify a coefficient for each semantic transference and refer to the two terms that are involved in this transfer. Then, in counting the frequency of the words of these sentences, the coefficient is multiplied by the frequency. The net PSO algorithm provides optimal clustering solutions. To increase the speed and precision of the system, we use two local searches based on the composition particle structure and, at the end, we see several percent improvement over the previous work.