Web hierarchy modeling

摘要

The present invention relates to systems, apparatus, and methods of modeling a hierarchical object relationship in a hypertext mark-up language (HTML) context. In some embodiments, a probabilistic determination is made of parentage of content objects being requested over a communications system. In certain embodiments, the content objects are intercepted by a proxy server as part of a prefetching system. When parentage is uncertain, a system may identify a plurality of nodes that are potential parent/root nodes for the content object, associate a likelihood with each parent/root node, and establish a potential parent/root node as the parent for the purposes of a prefetching. Various embodiments use referrers where reliable, supplemented or supplanted by other techniques (e.g., “rootable” URL determinations) to model the object hierarchies and parent/root relationships.

主权项

1. A method for prefetching at a proxy server based on root node identification for a requested HTTP object at the proxy server, the method comprising:
receiving a request for an HTTP object; determining, using a computing system, a plurality of candidate root nodes for the requested HTTP object, each candidate root node comprising an object that may have caused the request for the HTTP object; for each candidate root node:
determining a likelihood that the respective candidate root node is the root node that caused the request for the HTTP object, andassociating, at the computing system proxy server, the determined likelihood with the candidate root node; selecting one of the candidate root nodes from the plurality of candidate root nodes based on the determined likelihoods for each of the candidate root nodes; and establishing the selected candidate root node as the root node for the requested HTTP object.