Identification of effective parameters in summarizing Persian texts using GMDH neural networks

One of the most important challenges in search engines is to provide a sustainable way of finding texts related to a document. Depending on thecharacteristics of the Persian language, it is difficult to identify the pattern of words used and key. In this paper, a cluster-based approach ispresented to summarize multiple document texts. Two clustering strategies have been used to group efficiently and appropriately the sentences.Which is the use of limited single genetic clustering for clustering sentences, and the other is the automatic production of correlation vectors andvector word vectors. In this method, the GMDH numerical data grouping is used to determine the similarity between the sentences. The resultsshow that Which can be used to limit the hierarchical clustering of the neural network, Provide the best quality than the previous ones

Keywords:

Multi-Systematic Synthesis , Vector Definition , Clustering , Limited Hierarchical Clustering , GMDH Neural Network