Accurate text classification through selective use of image data

An online shopping system comprising:

• an index of documents;
• a search engine for receiving a query and returning a set of search results responsive to the query based on the index of documents;
• and
• a ranking engine for ranking the search results based at least in part on a classification based on the concatenation of a first distribution and a second distribution, wherein the first distribution and the second distribution result from inferences on different sets of training data using different classifiers.

A computer-readable storage device comprising computer readable instructions for classifying each search result from among a set of search results generated by a search engine, the computer readable instructions comprising instructions that:

• train a first classifier on a first set of training data, and infer a first distribution on a second set of training data;
• train a second classifier on the first set of training data, and infer at least one image-based distribution on the second set of training data;
• concatenate the first distribution with the second distribution;
• train a combined classifier based on the concatenation;
• classify each result from among a set of search results generated by the search engine;
• and
• rank the set of search results based at least in part on the classification.