In this paper an extensive similarity matching framework between moving object trajectories is examined to incorporate both common and misinterpreted‐hidden conditions accommodating various decision making applications. Trajectories, indicative of the behaviour of their corresponding moving objects, include diverse detail and complexity and therefore normalization under the self‐organizing map neural network formulates them in a comparable state. Then, a series of ‘comparison spaces’ are defined, where selected rules force the trajectories to inherit a pre‐specified form corresponding to the application needs. The process includes pure geometric, translation‐scale‐rotation invariant, topologic, conceptual and history‐based similarity matching under a holistic and scaled fashion. Both comparisons between pairs of trajectories and upon profile libraries are considered, while entire or partial matching is supported.