Augmented Search Queries Using Knowledge Graph Information
What are Augmented Search Queries?
Last year, I wrote a post called Quality Scores for Queries: Structured Data, Synthetic Queries and Augmentation Queries, which told us that Google may look at query logs and structured data (table data and schema data) related to a site to create augmentation queries, and evaluate information about searches for those comparing them to original queries for pages from that site, and if the results of the augmentation queries do well in evaluations compared to the original query results, searchers may see search results that are a combination of results from the original queries and the augmentation queries.
Around the time that patent was granted to Google another patent that talks about augmented search queries was also granted to Google, and is worth talking about at the same time with the patent I wrote about last year. It takes the concept of adding results from augmented search queries together with original search results, but it has a different way of coming up with augmented search queries, This newer patent that I am writing about starts off by telling us what the patent is about:
This disclosure relates generally to providing search results in response to a search query containing an entity reference. Search engines receive search queries containing a reference to a person, such as a person’s name. Results to these queries are often times not sufficiently organized, not comprehensive enough, or otherwise not presented in a useful way.
Augmentation from the first patent means possibly providing additional information in search results based upon additional query information from query logs or structured data from a site. Under this new patent, augmentation comes from recognizing that an entity exists in a query, and providing some additional information in search results based upon that entity.
This patent is interesting to me because it takes an older type of search – where a query returns pages in response to the keywords typed into a search box, with a newer type of search, where an entity is identified in a query, and knowledge information about that entity is reviewed to create possible augmentation queries that could be combined with the results of the original query.
The process behind this patent can be described in this way:
In some implementations, a system receives a search query containing an entity reference, such as a person’s name, that corresponds to one or more distinct entities. The system provides a set of results, where each result is associated with at least one of the distinct entities. The system uses the set of results to identify attributes of the entity and uses the identified attributes to generate additional, augmented search queries associated with the entity. The system updates the set of results based on one or more of these augmented search queries.
A summary of that process can be described as:
- Receiving a search query associated with an entity reference, wherein the entity reference corresponds to one or more distinct entities.
- Providing a set of results for the search query where the set of results distinguishes between distinct entities.
- Identifying one or more attributes of at least one entity of the one or more distinct entities based at least in part on the set of results.
- Generating one or more additional search queries based on the search query, the at least one entity, and the one or more attributes.
- Receiving an input selecting at least one of the one or more additional search queries and providing an updated set of results based on the selected one or more additional search queries, where the updated set of results comprises at least one result not in the set of results.
The step of generating one or more additional search queries means ranking the identified one or more attributes and generating one or more additional search queries based on the search query, the at least one entity, the one or more attributes, and the ranking.
That ranking can be based on the frequency of occurrence.
The ranking can also be based on a location of each of the one or more attributes with respect to at least one entity in the set of results.
This process can identify two different entities in a query. For instance, there were two versions of the Movie, the Planet of the Apes. One was released in 1968, and the other was released in 2001. They had different actors in them, and the second was considered a reboot of the first.
When results are generated in instances where there may be more than one entity involved, the search queries provided may distinguish between the distinct entities. They may identify one or more attributes of at least one entity of the one or more distinct entities based at least in part on the set of results. Augmented search queries may be generated for “one or more additional search queries based on the search query, the at least one entity, and the one or more attributes.”
This patent can be found at:
Providing search results using augmented search queries
Inventors: Emily Moxley and Sean Liu
Assignee: Google LLC
US Patent: 10,055,462
Granted: August 21, 2018
Filed: March 15, 2013
Methods and systems are provided for updating a set of results. In some implementations, a search query associated with an entity reference is received. The entity reference corresponds to one or more distinct entities. A set of results for the search query is provided, and the set of results distinguishes between distinct entities. One or more attributes for at least one entity of the one or more distinct entities are identified based at least in part on the set of results. One or more additional search queries are identified based on the search query, the at least one entity, and the one or more attributes. An input selecting at least one of the additional search queries is received. An updated set of results is provided based on the selected additional search queries. The updated set of results comprises at least one result not in the set of results.
Some Additional Information About How Augmented Search Queries are Found and Used
A couple of quick definitions from the patent:
Entity Reference – refers to an identifier that corresponds to one or more distinct entities.
Entity – refers to a thing or concept that is singular, unique, well defined, and distinguishable.
This patent is all about augmenting a set of query results by providing more information about entities that may appear in a query:
An entity reference may correspond to more than one distinct entity. An entity reference may be a person’s name, and corresponding entities may include distinct people who share the referenced name.
This process is broader than queries involving people. We are given a list in the patent that it includes, and it covers, “a person, place, item, idea, topic, abstract concept, concrete element, other suitable thing, or any combination thereof.”
And when an entity reference appears in a query, it may cover a number of entities, for example, a query that refers to John Adams could be referring to:
- John Adams the Second President
- John Quincy Adams the Sixth President
- John Adams the artist
In addition to having an entity in an entity reference in a query, we may see a mention of an attribute for that entity, which is “any feature or characteristic associated with an entity that the system may identify based on the set of results.” For the John Adams entity reference, we may also see attributes included in search results, such as [second president], [Abigail Adams], and [Alien and Sedition Acts].
It sounds like an entity selection box could be shown that allows a searcher to identify which entity they might like to see results about, so when there is an entity in a query such as John Adams, and there are at least three different John Adams that could be included in augmented search results, there may be clickable hyperlinks for entities for a searcher to select or deselect which entity they might be interested in seeing more about.
Augmented Search Queries with Entities Process Takeaways
When an original query includes and entity reference in it, Google may allow searchers to identify which entity they are interested in, and possibly attributes associated with that entity. This really brings the knowledge graph to search, using it to augment queries in such a manner. A flowchart from the patent illustrates this process in a way that was worth including in this post:
The patent provides a very detailed example of how a search that includes entity information about a royal wedding in England might be surfaced using this augmented search query approach. That may not be a query that I might perform, but I could imagine some that I would like to try out. I could envision some queries involving sports and movies and business. If you own a business, and it is not in Google’s knowledge graph you may end up missing out on being included in results from augmented search queries.
Copyright © 2019 SEO by the Sea ⚓. This Feed is for personal non-commercial use only. If you are not reading this material in your news aggregator, the site you are looking at may be guilty of copyright infringement. Please contact SEO by the Sea, so we can take appropriate action immediately.
Plugin by Taragana
The post Augmented Search Queries Using Knowledge Graph Information appeared first on SEO by the Sea ⚓.
About the Author - CBPO
- PageRank 2020
- Twitter plans to expand its misinformation labels — but will they apply to Trump?
- Sony’s ZV-1 compact camera zooms in on vloggers
- Extra Crunch Live: Join Verizon CEO Hans Vestberg for a live Q&A today at 2pm EDT/11am PDT
- Create Google Display Countdown Ads Using Rules: Getting Creative With Ad Customizers