The spam email, unwanted digital correspondence on the internet, can target the individual, group and or organization. Because email addresses provided for registration in online sign ups can be misused by devious off third parties (spammers), this type of communication is very risky once logged in by user. Disposable email addresses are practiced more frequently by spammers; by sending emails to such addresses, ones can access the messages only for a short while. This is a common method to send spam and that spammers often use their account information to send spam instead of disclosing who they really are. The outcomes of these attacks are severe, ranging from theft of the user login information to system overload and reduction in storage capacity. As a result, it is quite necessary to manage a suitable detection system that pinpoints using feature extraction and classification techniques to identify spam emails and temporary email addresses. An innovative approach based on the use of an application of a new Natural Language Processing based Random Forest (NLP-RF) technique is a promising approach. The purpose of the proposed method takes advantage of NLP (natural language) to improve the spam email filtering with the pattern of natural language and improves the spam detecting process. Integrating these two approaches increases the accuracy of spam email identification and reduce the prevalence of spam in the entire system.