Documents that may be returned in response to a query may include documents such as "web pages, news articles, image results, blog documents, etc." The result identification aspect of this process where entities are identified can provide information about the attributes of those entities. Entities may be ranked based on factors that determine their score. The first is when an entity is extracted from a document that has a less relevant topic. "For example, the first entity related to the movie" Toy Story 3 "is extracted from the document of the topic related to" Soccer ", and the second entity related to the movie" Inception "is the document of the topic related to" Soccer ". Suppose it is extracted from. movie. "
The entity ranking engine 230 determines ghost mannequin effect that the first entity is less relevant to the result of the first entity being extracted than the second entity is more relevant to the result of the second entity being extracted." can do. advertisement Continue reading below The terms in the document from which the entity was extracted may be reviewed. The IR (Information Retrieval) score of the document from which the ranked entity is extracted is used to determine the entity score used to rank that entity based on how relevant the page is to the query. You can decide. Therefore, an entity ranked as "Best Sci-Fi Novel of 2020" from the page with the highest IR score of the query, "Best Science Fiction Novel of 2020" may have a higher entity score than the entity on the page. there is. The IR score of "Several OK Science Fiction Novels in 2020" is high.
If the query for a ranked entity contains certain attributes such as publishing location, and the query is something like "2020's Best American Science Fiction Novel", then the ranked entity identifies the publishing location for that entity. It may be. As stated in the patent, the range of attributes used can be quite wide. advertisement Continue reading below "For example, suppose a particular entity is associated with the" Movies "category. In this example, the attributes may include information such as release date, leading actor and / or leading actress, supporting actor and / or actress, box office gross, and production director credits. Music credits, synopses, and / or summaries related to the movie. In another example, entities associated with different categories can be associated with different sets