Category constrained queries
View Patent ↗A user may submit a search query to a search engine. The search engine may process the search query and generate a set of results. Each of the items searched by the search engine may have been pre-assigned to a category in a category tree. Previous interactions by other users with the items after similar queries may have been recorded. The search engine may identify categories based on the distribution of the interacted-with results among the categories. The category tree may be analyzed at different levels, based on the entropy observed at each level. A level with low entropy may be chosen, and categories at that level used to constrain the query.
1. A system comprising:
a communication module, implemented using on or more processors, that is configured to receive, over a network from a client machine, a query;
a category identification module, implemented using on or more processors, that is configured to:
identify a first plurality of categories responsive to the query based on prior user interactions after similar queries;
identify an entropy value based on a distribution of the prior user interactions over the first plurality of categories; and
modify the first plurality of categories based on the entropy value to identify a second plurality of categories;
a constraint module, implemented using one or more processors, that is configured to modify the query to include the second plurality of categories as constraints; and
a query module, implemented using one or more processors, that is configured to execute the query to identify search results, the query including the second plurality of categories as constraints, the communication module configured to communicate the search results, over the network, to the client machine.
2. The system of claim 1 , wherein the first plurality of categories includes categories within a tree of categories.
3. The system of claim 2 , wherein the category identification module is further configured to modify the first plurality of categories, based on the entropy value in excess of a minimum threshold, to generate the second plurality of categories by replacement of at least one category of the first plurality of categories with a parent of the at least one category in the tree of categories.
4. The system of claim 1 , wherein the prior user interactions include a first prior user interaction and a second prior user interaction and wherein the first plurality of categories includes a first category and a second category.
5. The system of claim 4 , wherein the first prior user interaction includes purchasing an item that is described by the first listing that is included in the second category.
6. The system of claim 4 , wherein the second prior user interaction includes viewing an item that is described by the second listing that is included in the second category.
7. The system of claim 4 , wherein the first prior user interaction has a weight that is greater than the second prior user interaction.
8. A method comprising:
receiving, over a network from a client machine, a query;
identifying a first plurality of categories responsive to the query based on prior user interactions after similar queries;
identifying an entropy value based on a distribution of the prior user interactions over the first plurality of categories; and
modifying the first plurality of categories based on the entropy value to identify a second plurality of categories;
modifying the query to include the second plurality of categories as constraints;
executing the query to identify search results, the query including the second plurality of categories as constraints; and
communicating the search results, over the network, to the client machine.
9. The method of claim 8 , wherein the first plurality of categories includes categories within a tree of categories.
10. The method of claim 9 , wherein modifying the first plurality of categories comprises:
based on the entropy value exceeding a minimum threshold, modifying the first plurality of categories to generate the second plurality of categories by replacing at least one category of the first plurality of categories with a parent of the at least one category in the tree of categories.
11. The method of claim 8 , wherein the prior user interactions include a first prior user interaction and a second prior user interaction and wherein the first plurality of categories includes a first category and a second category.
12. The method of claim 11 , wherein the first prior user interaction includes purchasing an item that is described by the first listing that is included in the second category.
13. The method of claim 11 , wherein the second prior user interaction includes viewing an item that is described by the second listing that is included in the second category.
14. The method of claim 8 , wherein the identifying the entropy value includes generating the entropy value based on the first prior user interaction having a weight that is greater than the second prior user interaction.
15. A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
receiving, over a network from a client machine, a query;
identifying a first plurality of categories responsive to the query based on prior user interactions after similar queries;
identifying an entropy value based on a distribution of the prior user interactions over the first plurality of categories; and
modifying the first plurality of categories based on the entropy value to identify a second plurality of categories;
modifying the query to further include the second plurality of categories as constraints;
executing the query to identify search results, the query including the second plurality of categories as constraints; and communicating the search results, over the network, to the client machine.