Identifying and ranking attributes of entities

摘要

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for query analysis. Queries are identified in query data, and an entity-descriptive portion and a suffix are determined in each query. Query counts are determined for a number of times that the respective queries occur in the query data. Based on the query counts, an entity-level count is estimated, which represents a number of query submissions that include the particular suffix and are considered to refer to a first entity. The entity is determined to be a particular type of entity. A type-level count is determined, which represents a number of query submissions that include the first suffix and are estimated to refer to entities of the particular type of entity. A score is assigned to the particular suffix based on the entity-level count and the type-level count.

主权项

1. A method performed by data processing apparatus, the method comprising:
identifying queries in query data; determining, in each of the queries, (i) an entity-descriptive portion that refers to an entity and (ii) a suffix; determining query counts of a number of times that the respective queries were submitted; for at least a particular query of the identified queries, distributing the query count for the particular query among multiple different entities by assigning, to each of the multiple different entities, a partial query count that is an estimate of a number of submissions of the particular query that refer to the entity; estimating, based at least in part on one or more of the partial query counts, an entity-level count of query submissions that include a particular suffix and are considered to refer to a first entity of the multiple different entities; determining that the first entity is a particular type of entity; determining a type-level count of the query submissions that include the particular suffix and are estimated to refer to entities of the particular type of entity; and assigning, based on the entity-level count and the type-level count, a score for the particular suffix with respect to the first entity.