The previous studies in this field have focused on content and meta-content based features. The main assumption in this area of spam detection of late is that human web usage behavior is intrinsically different from spambot behavior. The tools used for this purpose are generally based on challenge-response techniques. This is attributed to the reason that spambots have some different malicious intentions. Spambots visit the website mainly to spread spam content rather than to consume the content. Hence by investigating and mining web usage data it is possible to distinguish spambots from human users. This proposed field of research is a rule based on-the-fly web spambot detection method. he method is based on web usage behavior. Discriminative features called action strings from web usage data are extracted to classify spambot vs. human. An action as a set of user efforts to achieve certain purposes and action strings as a sequence of actions for a particular user in a transaction are proposed.