Distinguishing accessories from products for ranking search results

A computer implemented method comprising:

• receiving, by an online search engine including at least one sever and at least one data storage device, a search query that was submitted to the search engine by a user device;
• identifying, by the search engine and using the received search query, ranked search results that include first results of a first type and second results of a second type, wherein a particular result of the second type has an initial rank position in the ranked results;
• determining a first likelihood that each of one or more of the ranked results is a first type of result based on an attribute of the ranked search result and an attribute distribution for items of the first type that are responsive to the query;
• determining a second likelihood that each of one or more of the ranked results is a second type of result based on the attribute of the ranked search result and an attribute distribution for items of the second type that are responsive to the query;
• determining, based on the first likelihood and the second likelihood for each of the ranked results, a first number of first results that are included in the ranked results and a second number of second results that are included in the ranked results;
• classifying, by the search engine, the query as a query requesting results of the first type based, at least in part, on the first number of first results exceeding the second number of second results;
• re-ranking, by the search engine, the ranked results based on the classification of the query, wherein the re-ranking lowers the rank position of the particular result relative to the initial rank position of the particular result;
• and
• providing, to the user device and in response to the query, a search results page that presents search results specifying at least a subset of highest ranked results in the re-ranked results.

A system, comprising:

• a search engine including at least one server;
• and
• machine readable instructions stored on a computer readable medium, wherein execution of the instructions cause the search engine to perform operations comprising:
• receiving a search query that was submitted to the search engine by a user device;
• identifying, using the received search query, ranked search results that include first results of a first type and second results of a second type, wherein a particular result of the second type has an initial rank position in the ranked results;
• determining a first likelihood that each of one or more of the ranked results is a first type of result based on an attribute of the ranked search result and an attribute distribution for items of the first type that are responsive to the query;
• determining a second likelihood that each of one or more of the ranked results is a second type of result based on the attribute of the ranked search result and an attribute distribution for items of the second type that are responsive to the query;
• determining, based on the first likelihood and the second likelihood for each of the ranked results, a first number of first results that are included in the ranked results and a second number of second results that are included in the ranked results;
• classifying the query as a query requesting results of the first type based, at least in part, on the first number of first results exceeding the second number of second results;
• re-ranking the ranked results based on the classification of the query, wherein the re-ranking lowers the rank position of the particular result relative to the initial rank position of the particular result;
• and
• providing, to the user device and in response to the query, a search results page that presents search results specifying at least a subset of highest ranked results in the re-ranked results.