A web robot is an automated program that requests web resources independently from a web server. The growth of bad and good web robot traffic on web 2.0 over the years is on the high side compared to humans. A malicious web robot has been used to spam activities in web 2.0. It, however, poses serious challenges to website owners and can cause Distributed Denial of Service (DDoS) attacks. Previous web robot detection techniques emphasized pre-processing access-log files for identifying web robot session. A few approaches considered the semantic analysis of the content requested within a session as a source of web robot detection. Furthermore, very little effort has been made in combining the strength of behavioural features and semantic features in web robot detection. Therefore, this paper aimed at developing hybridized spambot detection system using source-content classification. This paper revealed that 7568 unique sessions were identified with 6993 humans, 558 spambots and 17 non-spambots; session text coherence (STC), session word relatedness (SWR) and session topic coherence (ST) were functionally expressed as = /( * ), = /( * ), and = /( * ) respectively, where n represents the number of relevant topics, m number of top words in each topic, sw sum of weights, c count of unique topics and k count of unique word. The hybridized features performed effectively with an accuracy of 93.67% compared to the individual features.