Automatic recognition of text keywords is one of the most widely used branches of Natural Language Processing Science today. The keywords of a text contain its main topics and are of great value to researchers and scientists. In this tool, we have used a comprehensive method which is a combination of observer-based and non-observer-based methods. In this method, first, the list of candidate keywords is extracted and after determining the characteristics of each of them, the keywords are sorted based on a very strong ranking method. This ranking method has doubled the results by normalizing the characteristics and genetic algorithm. The results of our experiments on a body of Persian news texts show a rate of 28.8% of criterion f.