If the site owner wants to share information on the web using spreadsheets or databases, they can do so by posting the data online and allowing others to access it. I also wrote a post about this topic on Natural Language Query Responses. We see Google with patents focused on answering natural language queries using data tables. This suggests that Google is working on ways to answer natural language queries using data tables. The blog post today is about how to answer questions about spreadsheets or databases that are written in natural language.
A searcher may want to access information in many different ways. This is something that Google has patented. I will review the patent granted in May 2021 and summarize its content.
How The Search Engine Might Make It Easier For a Searcher to Use Data Tables in Spreadsheets
The patent begins with data tables in spreadsheets and then demonstrates how the search engine could make it simpler for a searcher to utilize that information.
A spreadsheet is a document that contains tabular data organized into rows and columns. This data can be stored under different categories.
Sometimes the spreadsheet may perform calculation functions. A searcher may want certain data from a spreadsheet. The searcher can then construct a search query to look for the desired data they want in a database.
The searcher-desired data set may not always be stored in the spreadsheet. The user can use the data already present in the spreadsheet to get the desired information.
The person looking for information may go over the spreadsheet to find relevant data to use in a formula to get a result using the spreadsheet’s calculation function.
For example, if you have a spreadsheet with test scores for each student in a class, you may want to know the average score of the class.
After that, the person who did the search may have to calculate the average score for the class by adding together all of the test scores and then dividing that sum by the total number of students.
The data table may then calculate the average score of the class based on the preset formula.
In order to calculate the desired information, the searcher may need to input a pre-determined formula into the data table. This can be inefficient when dealing with large amounts of data. The searcher must have a deep understanding of how databases work in order to be effective.
The Patent Covers Systems and Methods to Process Natural Language Queries on Data Tables
The patent granted covers systems and methods to process natural language queries on data tables, such as spreadsheets. This will help the user to get information from the data table more easily.
The patent states that natural language queries may come from a searcher.
A query term and a grid range from a data table may be found relevant to the query term.
A summary of the data in a table may include entities based on the range of the data in the table.
A logic operation may be used to determine the number of data entities that match the query term.
The logic operation may be converted into a formula that can be executed on the data table. After you have input the data and the formula, you can ask the software a question in natural language and it will generate a result.
Natural language queries can be entered into a search engine in various ways, including through a searcher interface on a computer, or by entering the query manually or vocally.
You can also send queries to servers from computers through HTTP requests.
Those Natural Language Queries Can Originate In a First Language
Queries in a first language can be translated into a second language for processing.
A grid range is a computer that is not a server or a data table. It receives natural language queries and is located at the server.
The data table is a table of data stored on a computer, a remote server, or in a cloud.
The number of data entities can be dimensions, dimension filters, or metrics.
A searcher will see the results of their search in a visualization format, which can include an answer statement, a chart, or a data plot.
The search engine may get feedback from searchers after it provides them with results. When someone enters a query and the searcher feedback is positive, the formula is associated with that query.
An alternative interpretation of natural language queries may be provided when the searcher feedback is negative. An alternative result may be provided based on an alternative interpretation.
A Searcher Can Enter a Query For Data in Natural Language
Systems and methods that allow a searcher to enter a query for data in natural language are provided. The searcher can enter a query in natural language, and the system can process the query to identify relevant data.
The user’s natural language query is converted into a structured database query. If the structured database query shows that the data you want is not available in the data table, you may be able to find some related data in the table that could help you generate the data you need. Next, the formula can be automatically generated to calculate the desired information using the available data.
In other words, if you have a table of data like a list of test scores for every student in a class, you can use natural language processing to query that data. What is the average score of the class?
The words in the natural language query are extracted and used to interpret what is being asked. For example, the words “what,” “is,” “the,” “average,” “score,” “of,” “the,” and “class” are taken from the query to determine what is being asked.
The term “average score” may be a key term of the query based on the terms that are commonly used.
There is no data entry available in the spreadsheet for the “average score” category. For example, there is no column header for “average score.”
Logic can create an average score from the data that is already present.
An average score may be calculated by adding up all the class test scores and dividing the sum by the total number of students.
Then, a formula can be automatically generated to calculate the “average score” based on the natural language query, and the calculation result can be outputted to the searcher.
A formula can be generated that is associated with a tag, like “average score.” So even if more data is added to the spreadsheet – like new test scores for more students – the formula can still be used to automatically calculate an average score for the class in response to a natural language query.
How A Searcher May Get an Answer About Their Data
A person looking for information about their data may be able to find it more quickly and efficiently by using a search engine, rather than entering formulas or doing other types of analysis by hand.
The search engine may help users who are not familiar with all the features of spreadsheets to create organized queries or formulas.
The system has a server that communicates with a remote database and/or searcher devices over a network. The system may also include other related entities. The devices have interfaces for searching (generally called searcher interfaces).
Every device used for searching has its own personal computer, laptop, tablet, smartphone, or other type of communication device.
When searchers use the search engine, they are accessing and receiving information from the server and remote databases that are connected to the internet.
The device may have an input device to input data and an output device to output results.
How The Searcher Device Works
The user types a question into the search bar, and the computer processes it.
The device may process a natural language query and search within a local database.
The query searcher may also send the natural language query to a server that is able to store data tables and use a processor to analyze the query.
This means that the server can send and receive information, and access external databases for more data if necessary.
This means that when you enter a query in a natural language, it is translated into a database query, which can then be executed locally on your device, on the data tables stored on the server, or on remote databases (e.g. in the cloud, etc.).
The searcher device may have an application installed that allows the searcher to review data and enter a natural language query.
Geolocation
often relying only on either a Google account or browser settings doesn’t give Google’s algorithm enough confidence in the desired language of a query To increase certainty, they will track the user’s location.
Google uses a user’s physical location to target search results. A user in the US that searches for “Giants” will see more New York Giants results on the first Google results if they are on the East Coast of the United States – even during the NFL off-season. A West Coast user will see more San Francisco Giants results on the first Google results – even during the MLB off-season.
For many searches, there will not be a large difference in the results on Google.com from various locations. However, some searches will see significant changes. A search for the term “football” will be largely the same in the United States, Canada, and the United Kingdom, while a search for the term “holiday” will be noticeably different in the United Kingdom than it is in the United States.
TLD of Google Domain
The user’s physical location is not always the best indicator of their language intent. Other factors such as account or browser settings may be more accurate. The Google TLD that the query is conducted on can override the settings.
Typically, a logged-in user will see the Google.com homepage even if they are traveling outside the US. Users who are not logged in will be redirected to the local Google domain based on their browser settings, regardless of their preference for English or the US.
The Top Level Domain (TLD) is a very important factor in determining which language to return results in. If there was a hierarchy in Google’s language determination processing, it could either be first or simply go hand-in-hand with location targeting. The Top Level Domain (TLD) can give Google a good indication of the user’s language intent if the user has deliberately chosen a specific TLD.
A user who conducts a search on Google.com.br is very likely interested in Portuguese results. What this is saying is that if you are trying to use the TLD to figure out where the user is from, it might not be a good idea if the user is just traveling and not actually from that location. If you are an American traveling in Germany and you do a Google search while not logged in to your account, you will see Google.de by default because of your location. If Google only relies on the top-level domain (TLD) to determine a website’s language intent, it might give users inaccurate results.
If someone searched for the word “handy” in Germany, they would see results related to mobile phones because this is what Germans use to refer to a cell phone. The user would have seen different results if they had picked the correct language option.
Google always assumes the primary language of a country when using TLD for language. A search for the word “baguette” in Canada would return English results even though it is technically a French word. Google would assume that a query is in German in Switzerland even if German, French, and Italian are widely spoken.
Query Parsing and Matching
finally, Google analyzes the word itself to see if it can find any clues about what language it is. The algorithm looks for matches for the word in the most common languages. If a language is matched using a keyword, any resulting content will probably be in that language. This is fairly simple when the word is spelled correctly and only matches a single common language. When it’s not an exact match, it’s a bit more complicated.
In these cases, Google will look for things like matches between a misspelling and another word in a specific language. Google will try to use all the other clues to determine if the user made a spelling mistake or whether results in another language were actually sought. If you’re interested in the technical details of this process, you can read Google’s patent on the topic.
TLDR
SEO’s typically focus on the aspects of Google’s algorithm that decide how high up a webpage should be positioned in search results. Although it may appear that Google’s algorithm for ranking content is simply a score-based system, it is actually much more complex. The search engine needs to analyze each query in real time to figure out the user’s language before it can start retrieving sites from its index and determining the ranking for each page.
After reading this, it should be clear how much effort Google put into determining the language of a query so they can provide the best possible results. I have been unable to find any evidence from Google which details how they establish ranking, and the information above is based on my own investigation. I’m always interested in learning about new things. So if you have any information that you think I would find interesting, please share it with me.
Related posts:
