Google Universal Search 2200 In a Digital Assistant
On October 1, 2019, I wrote about a continuation patent update to Google’s Universal Search Results. It was the fourth time that particular patent had gotten updated. I wrote about it under the post name – Google’s New Universal Search Results. It makes sense to see what new changes have happened with a new continuation patent in Google Universal Search 2200.
The inventors listed on the patent are the same as on the original version of the patent, and the description looks similar. The changes are to the claims section of the patent, which the prosecution officers at the USPRO look at when deciding whether to grant or deny a continuation patent.
A continuation patent will assume the patent’s filing date or patents that it continues, with the claims section showing off new processes in what the patent excludes others from doing.
So ideally, we should compare the newest claims and the older ones. I have reached the first claims from the first four patent versions. What news language or processes are in this fifth version of the patent? Here is the first version:
United States Patent 11,314,822
Granted: April 26, 2022 – Interface for a universal search
Inventors: Bret S. Taylor, Marissa Ann Mayer, Orkut Buyukkokten
A search engine may search for a user search query over several possible search categories. For example, the search query may be performed for general web documents, images, and news documents. The search engine ranks categories based on the search query and the records returned for each type and presents the search results to the user by sort. Higher ranking categories may be given more prominently than lower-ranking categories.
Here is the first claim of this latest version of Google’s Universal Search Patent:
What is claimed is:
1. A computer-implemented method comprising:
- Receiving a search query based on voice recognition of user input submitted through a digital assistant
- Determining a relevance score for each particular category of responses to the search query among multiple different categories of responses to the search query, the relevance score for each category indicates the relevance of responses to the search query that corresponds to the particular category
- Providing, in response to the search query, an output that specifies relevant categories of responses to the search query, including
- Organizing presentation of the output such that the multiple different categories get provided in a specified order according to the relevance scores that indicate the relevance of each of the particular categories concerning the search query
- Including, in the output for each of two or more categories, search results relevant to each of the two or more different categories.
I am also going to post the first claim of the earlier version of the Google Universal Search patent to make the two easier to compare:
The First claim from the last continuation version os Google’s Uuniversal Search patent (May 17, 2016) looks like this:
1. A computer-implemented method comprising:
- Receiving a search query
- Obtaining data indicating multiple sets of resources responsive to the search query, each of the various stages of resources getting classified as corresponding to a different category of resources
- Determining a relevance score for each of the different categories, the relevance score for each category indicates the relevance of the set of resources corresponding to the class concerning the search query
- Providing, in response to the search query, a search results page that presents search results identifying resources from the multiple sets of resources, including
- Organizing the search results for display such that search results corresponding to different categories get provided in respective areas that have locations determined according to the relevance scores for the various types that indicate the relevance of the individual sets of resources concerning the search query
- Including, in the respective areas provided for each of two or more of the different categories, search results that are relevant to the two or more different categories, wherein a second-most appropriate variety has fewer search results
- The search results page presents a most relevant category based on the most relevant category is more relevant to the search query than the second most relevant category
What is New in Google Universal Search 2200?
1. The new claims refer to a “search query based on voice recognition of user input submitted through a digital assistant.” This claim is the first reference to voice recognition or a digital assistant.
2. The older claim specifies that the number of search results shown to a searcher has fewer search results than earlier categories of search results that might appear above it. The most recent version of the claims doesn’t contain the “fewer” results from the results that appear above it. Why is this important? The patent doesn’t say, but it seems an organizing idea behind the patent now.
3. Again, this patent doesn’t list the different types of results in SERPs in response to a query. A claim from the 2008 version of the patent told us specifically about the effects in Universal Search results: “The method of claim 1, where the document categories include at least one of a news category, an image category, or a product category.” That is the last time that Google has listed different types of results that will appear in Universal search results.
5. There is more language about voice recognition in the claims for the newest version of the Google Universal Search 2200 patent. It tells us that: “11. The system of claim 8, wherein providing the output that specifies the relevant category of responses comprises providing the output through a speaker.” It can provide these search results through a speaker. So it appears that Google is updating this Universal Search Results 2200 patent to be helpful when being used as part of a digital assistant system.
This appears to fit in with the movement towards conversational search (Lamba) that Google has announced and search results that are stories and conversations, and more automated assistant patents from the search engine.