Dynamic categorizations for electronic advertising

摘要

An advertiser submitting a sponsored ad to a provider can determine aspects of the ad, such as creative elements and bid price, based on a dynamic categorization of an item and/or landing page associated with the ad. An algorithm can be used that utilizes a browse tree for each item to be displayed to a user following a sponsored link, for example, and attempts to find a node in the browse tree that appears at the lowest level in the tree and with a sufficient appearance frequency. When consensus is reached for a node in the browse tree that meets these criteria, a corresponding categorization is determined, which can be used to determine appropriate aspects to be used for the sponsored ad, at least until enough information is obtained such that aspects can be determined independent of the categorization.

主权项

1. A computer-implemented method, comprising:
obtaining a request to determine a category for a page from among a plurality of categories organized in a hierarchical category browse tree, each category of the plurality of categories corresponding to one of a plurality of category browse nodes in the hierarchical category browse tree, the page associated with one or more items, each of the one or more items associated with a respective-subset of the plurality of categories; determining a browse path for each of the one or more items to identify a plurality of browse paths through the hierarchical category browse tree based at least in part by traversing the hierarchical category browse tree from an initial category browse node to a category browse node corresponding to a respective category of the respective-subset of the plurality of categories associated with a respective one of the one or more items; determining ranking data associated with each of the plurality of category browse nodes; determining a respective weight for the plurality of category browse nodes based at least in part on the ranking data; for each level of the hierarchical category browse tree, determining a frequency of appearance for each category browse node of the plurality of category browse nodes that is included in one of the plurality of browse paths; determining a respective value for each of the plurality of category browse nodes based at least in part upon a respective frequency of appearance of a category browse node of the plurality of category browse nodes and the respective weight for each of the plurality of category browse nodes; and selecting, for the page, the category associated with a category browse node of the plurality of category browse nodes whose respective value meets a specified threshold.