Malicious (spam) social bots generate and spread fake tweets and automate their social relationships by pretending like a follower and by creating multiple fake accounts with malicious activities. Furthermore, malicious social bots post shortened malicious URLs in the tweet in order to redirect the requests of online social networking participants to some malicious and suspicious servers. Hence, distinguishing malicious social bots from legitimate users is one of the most tasks in the Twitter network. To detect malicious or suspicious social bots, extracting URL-based features that include frequency of shared URLs, DNS fluxiness feature, network features, link popularity features and spam content presents in URLrequires less amount of time comparatively with social graph-based features (which rely on the social interactions of users). Moreover, malicious social bots cannot quickly manipulate URL redirection chains. In this, a learning automata-based malicious social bot detection (LA-MSBD) algorithm is a Machine Learning approach proposed by integrating a Naïve Bayes algorithm model with URL-based features(URL Classification and Feature Extraction) for identifying trustworthy participants (users) in the Twitter network. Experimentation has been performed on 2 Twitter data sets, and the results obtained illustrate that the proposed algorithm achieves improvement in precision and detection accuracy. KEY WORDS: Learning Automata, Friends recommentation, URL Classification, Feature Extraction, Online social network.