It is not always easy for Google to interpret the relevant entities from the search queries. Search queries can be automatically enriched with additional semantic information or annotations in the background or suggested to the user via autosuggest. The matching of the search query and the entity no longer takes place solely on the basis of the text entered, but also takes into account semantic relationships between entities and attributes.
In the following example, the entity "Ann Dunham" is searched japan phone number data for. A purely term-based search engine would have problems answering the search query for "Obama's parents". Through the interaction of term-based and entity-based search, the search engine can return the answer "Ann Dunham" as the mother.
Source: Entity oriented Search, Krisztian Balog
In practice, the result looks like this. In addition to the mother Ann Dunham, the second entity searched for, the father Barack Hussein Obama Senior, as well as the knowledge panel of Barrack Obama are also displayed. The knowledge panel of Barrack Obama is displayed because the system reacts to the term Barack Obama in the search query. The other two entity boxes are displayed based on the additional semantic information from the knowledge graph.
The practical thing about this dual system when interpreting search queries is that results can also be returned if no entity is searched for in a search term.
In addition to the entity-based interpretation of search queries, the term-based method can also be supported by entity type-based methods. This is relevant for search queries in which several entities from a type class are searched for, such as "sights in Hanover". A box is displayed here that lists several entities.
Often, search queries relating to an entity type will display either the currently most relevant entity, carousels or the above type of knowledge graph boxes. In such boxes, the most relevant entities are displayed depending on their weighting in relation to the search query. Google can determine the proximity or relevance of the entities to the search query in a similar way to documents using vector space analyses such as Word2Vec . The smaller the angle between the search query vector and the entity vector, the more relevant or closer the term and entity are.
Example Vector Space Analysis Entities for the Search Query
refinement of search queries
By refining search queries, also known as query refinement, search engines can rewrite search terms in order to give the search query a more precise meaning and deliver corresponding search results. This can take place in the background without the user noticing, or it can be actively triggered by the user. Especially with search queries that are very generically aimed at an entity, it is often unclear what information the user wants to find. Up to now, the most important data about an entity has been displayed in a standardized manner in the Knowledge Panel, geared towards the respective entity type. (For more information, see my article How does Google create Knowledge Panel & Knowledge Cards? ). In the future, however, users will also be able to actively specify in more detail which media formats and information they would like to see about the entity using buttons, as screenshots from the American SERPs show. The search queries are obviously rewritten when the button is clicked.