Abstract: A method of processing a new visitor session to a web-based system, the method including: for each of a plurality of web requests made to the web-based system in the new visitor session, determining a request vector corresponding to the web request, wherein each request vector is derived from the learned parameters of a trained artificial neural network configured to determine a plurality of probabilities from an input web request or an input sequence of web requests, wherein each of the plurality of probabilities represents a likelihood that, within a session containing the input web request of sequence of web requests, a respective one of a plurality of possible web requests will be requested at one or more predetermined locations before or after the input web request or sequence of web requests; and combining the request vectors to obtain a session vector which characterizes the new visitor session.

Abstract: A method of processing web requests directed to a website includes, at a system for processing web requests: (i) for each of a plurality of web requests directed to a website, determining a request vector corresponding to the web request, wherein each request vector represents multiple predetermined features of the respective web request; (ii) clustering the request vectors by respectively assigning each request vector to one of a plurality of clusters using a clustering algorithm; (iii) repeatedly updating the clustering of request vectors using the clustering algorithm such that the plurality of clusters dynamically change over time; (iv) monitoring cluster metadata associated with each cluster as the plurality of clusters dynamically change over time, (v) identifying, based on monitoring, any cluster meeting a predetermined anomaly criterion indicating that the cluster is displaying anomalous behaviour; and (vi) triggering an investigation of a cluster identified as meeting the predetermined anomaly criter

Abstract: A method of processing web requests directed to a website, the method including: (i) receiving a plurality of web requests directed to the website; (ii) for each of the plurality of web requests, identifying a source from which the web request has originated; (iii) for at least one web request identified as having originated from a given source: determining whether the source is a bot or a non-bot based on the at least one web request; if the source is determined to be a bot, using a machine learning engine to assign one of a plurality of predetermined bot categories to the source based on the at least one web request.