Person-specific data graphs to assist queries and predictions

Share it’s love!

A latest Google patent is for question and forecast assist, specializing in consumer particular graphs of information.

These user-specific data graphs could also be particular to specific customers.

Which means that Google can use these graphs to offer leads to response to a number of queries submitted by the consumer, and / or to spotlight knowledge which may be related to the consumer.

I remembered one other patent that I wrote lately after I noticed this patent, within the article, Answering questions utilizing data charts, during which Google can search on a query requested by a consumer and create a data graph from the returned search outcomes. use to search out the reply to their query.

So, Google doesn’t have just one data graph, however can use lots of them.

Information for questions that may be requested, or for various folks asking these questions.

This user-specific data graph patent tells us that the modern points of the underlying course of embody:

  1. Receipt of user-specific content material
  2. Person-specific content material may be related to a consumer of a number of IT departments
  3. This user-specific content material is processed with the assistance of a number of analyzers to establish a number of entities and a number of relationships between these entities.
  4. An analyzer particular to a schema and a number of entities and a number of relationships between entities recognized primarily based on the schema
  5. This course of gives a number of user-specific data graphs.
  6. A user-specific data graph that’s user-specific, which incorporates nodes and edges between nodes to outline relationships between entities primarily based on the schema
  7. The method contains storing the consumer particular data graph (s).

Non-compulsory options involving the availability of a number of user-specific data graphs may additionally embody:

  • Figuring out {that a} node representing a number of entities and an edge representing a relationship related to the entity are lacking from a user-specific data graph
  • Including the node and edge to the user-specific data graph
  • The sting connecting the node to a different node of the user-specific data graph

The actions additionally embody:

  1. Obtain a request
  2. Receipt of a number of user-specific outcomes that reply to the question
  3. The consumer particular end result (s) are supplied primarily based on the consumer particular data graph (s).
  4. Present the consumer particular end result (s) for show to the consumer
  5. An edge is related to a weight
  6. The burden indicating the relevance of a relationship represented by the sting
  7. The worth of the burden will increase relying on the strengthening of the connection within the subsequent content material particular to the consumer
  8. The worth of the burden decreases relying on the dearth of reinforcement of the connection within the subsequent content material particular to the consumer
  9. A lot of user-specific data graphs are supplied primarily based on the user-specific content material.
  10. Every user-specific data graph is restricted to a respective schema
  11. Person-specific content material is supplied by way of using a number of companies applied by the pc by the consumer

Advantages of utilizing the user-specific data graph system

The patent describes the benefits of finishing up the method described on this patent:

  1. Supplies a structured solution to seize data about particular person customers
  2. Activation of leads to response to complicated queries, for instance a collection of queries, regarding a consumer
  3. The user-specific data graph can present a single canonical illustration of the consumer primarily based on the consumer exercise deduced from a number of computer-implemented companies. .
  4. Customers' actions might overlap, with the matching of the user-specific data graph making certain {that a} canonical entry is supplied for every exercise.
  5. The mix of those components might end in a common data graph, reminiscent of a non-user-specific data graph and user-specific data graphs.

(This common graph of information appears attention-grabbing.)

Data from sources reminiscent of these can be utilized to create user-specific data graphs:

  • Social community of a consumer
  • Actions or social actions
  • Job
  • Preferences of a consumer
  • Present location of a consumer

That is in order that the content material that is likely to be extra related to the consumer is utilized in these data graphs.

We’re additionally instructed that "the identification of the consumer may be dealt with in order that no personally identifiable data may be decided for the consumer", and that "the situation geographic location of a consumer may be generalized so {that a} specific location of a consumer cannot be decided ".

The user-specific data graph patent

This patent may be discovered at:

Structured consumer graphic to assist queries and predictions
Inventors: Pranav Khaitan and Shobha Diwakar
Recipient: Google LLC
US Patent: 10,482,139
Granted: November 19, 2019
Submitting date: November 5, 2013

Summary

The invention pertains to strategies, programs and equipment, together with pc applications encoded on a pc storage medium, for receiving user-specific content material, user-specific content material. being related to a consumer of a number of computer-implemented companies, processing the user-specific content material with the help of a number of parsers to establish a number of entities and a number of a plurality of entity relationships, a schema-specific analyzer, and a number of entities and a number of entity-identified relationships primarily based on the schema, offering a number of graphs of user-specific data, a graph of user-specific data and together with nodes and edges between nodes to outline relationships between schema-based entities, and storing e or the user-specific data graphs.

What’s the content material of the user-specific data graphs?

The forms of companies whose user-specific data graph data could possibly be retrieved might embody:

  • A search service
  • An electronic mail service
  • A chat service
  • A doc sharing service
  • An agenda sharing service
  • A photograph sharing service
  • A video sharing service
  • Weblog service
  • A microblogging service
  • A social community service
  • A location service
  • A registration service
  • An analysis and revision service

A user-specific data graph system

This patent describes a search system that features a user-specific data graph system that’s a part of this search system, instantly linked to or linked to the search system over a community.

The search system can work together with the user-specific data graph system to create a user-specific data graph.

This user-specific data graph system can present a number of user-specific data graphs that may be saved in an information retailer.

Every user-specific data graph is restricted to a consumer of a number of computer-implemented companies, for instance, search companies supplied by the search equipment.

The search system might work together with the user-specific data graph system to offer a number of user-specific search leads to response to a search question.

Structured Person Graphs for Queries and Forecasts

A user-specific data graph is created primarily based on the content material related to the consumer.

These user-specific data graphs embody quite a few nodes and edges between them.

A node represents an entity and an edge represents a relationship between entities.

The nodes and / or entities of a user-specific data graph could also be supplied primarily based on the content material related to a respective consumer, to which the user-specific data graph is restricted.

Graphs and schemas of information particular to the consumer

Person-specific data graphs may be created primarily based on a number of schemas (examples comply with). A schema describes how the info is structured within the user-specific data graph.

A schema defines a construction for the data supplied within the chart.

A schema constructions the info primarily based on domains, varieties, and properties.

A website contains a number of varieties that share a namespace.

A namespace is supplied as a listing of uniquely named objects, with every namespace object having a singular title or distinctive identifier.

For instance, a kind signifies a relation "is a" regarding a topic and is used to comprise a group of properties.

A topic can signify an entity, reminiscent of an individual, a spot or a factor.

Every of those subjects may be related to a number of varieties.

A property may be related to a topic and defines a "to 1" relationship between the topic and a worth of the property.

In some examples, the worth of the property might embody one other topic.

A user-specific data graph may be created primarily based on the content material related to a respective consumer.

This content material may be processed by a number of analyzers to tell the structured graph particular to the consumer.

An analyzer may be particular to a selected schema.

Belief or weight in relationships

The assigned weights between the nodes point out a relative pressure within the relationship between the nodes.

The weights may be decided primarily based on the content material related to the consumer, which content material underlies the availability of the user-specific data graph.

This content material can present a single occasion of a relationship between nodes or a number of cases of a relationship between nodes.

So there could be a minimal worth and a most worth.

Weights may also be dynamic:

  • Variation in time in keeping with the content material related to the consumer
  • Based mostly on the content material related to the consumer for the primary time
  • Based mostly on content material or lack of content material related to the consumer at a second time
  • Content material on the first time can point out a relationship between nodes
  • Weights can decompose over time

Multi-user particular data graphs

A number of user-specific data graphs may be supplied for a selected consumer.

Every user-specific data graph may be particular to a selected schema.

Usually, a user-specific data graph contains data a few particular consumer in a structured manner. (It represents a part of the world of the consumer by way of the content material related to the consumer through a number of companies.)

The data captured within the user-specific data graph might embody things like:

  • actions
  • Motion pictures
  • Meals
  • Social connections, for instance, in the actual world and / or digital world
  • Training
  • Basic likes
  • Basic disgust

Person-specific data graph versus user-specific social graph

A social graph comprises details about who an individual could possibly be linked to. A user-specific data graph additionally displays data about these connections, reminiscent of actions shared between individuals who may be linked to a data graph.

Examples of user-specific queries and graphs of information

Example of a user-specific knowledge graph

These are examples of the patent. Word that searches, emails, and social media postings can all work collectively to create a user-specific data graph, as proven within the mixed messages / actions under, taken collectively; which may result in a rise within the weight of the perimeters between the nodes and nodes and edges so as to add to this graph of information.

Pattern search question (taking part in tennis with my youngsters for the mountain) to a search service

Search Outcomes: Who can present details about taking part in tennis with kids in Mountain View, California.

Nodes may be supplied, a consultant of the "Tennis" entity, a "Mountain View" consultant, a "Household" consultant and a pair every representing a "Youngster".

It’s potential to offer an edge that represents a "/ Location / Play_In" relationship between the nodes, one other edge might signify a "/ Sport / Played_With" relationship between the nodes and different edges that will signify relationships. Household / Member_Of "between the node and the nodes.

Weights may be generated for every of the perimeters to signify completely different values.

An individual might publish the next instance: "We had fun taking part in tennis with our children in the present day!" In a social community service related to location-based knowledge indicating Mountain View, California .

Nodes may be recognized as representing tennis, mountain view, household and kids, in addition to the perimeters between these nodes.

Weights may be generated between these edges.

Somebody might obtain an e-mail from a resort stating "Affirm your resort reservation in Waikiki, whats up." October 15-20, 2014. We stay up for making your loved ones trip satisfying! "

Nodes may be added to the user-specific data graph, the place they signify the "Trip" and "Waikiki" entities.

Edges may be created within the user-specific data graph in response to this e-mail that represents a "/ Vacances / Travelled_With" relationship between the nodes, one which represents a "/ Holidays / CityTown" relationship between the nodes and one other edge that represents a "/ Trip / CityTown" relationship between the nodes.

Synchronization nodes may also be related to different nodes, reminiscent of a synchronization node representing October 2014 or a node representing a date vary from October 15, 2014 to October 20, 2014.

The consumer can submit the pattern request (kids's tennis course in Waikiki) to a search service.

Nodes may be created within the user-specific data graph representing tennis, Waikiki, household, and kids, in addition to respective edges between a minimum of a few of the nodes.

This pattern search question can improve the relevance of various entities and the relationships between the entities and the consumer involved.

This reinforcement can result in a rise within the respective weights related to the perimeters.

The consumer can obtain an electronic mail from a tennis membership, which can embody "Affirmation of Tennis Programs on the Tennis Membership, Waikiki, Hello".

The knots signify tennis and Waikiki, in addition to the perimeters that separate them.

This electronic mail enhances the relevance of the entities and the relationships between the entities for the actual consumer.

The weights between entities could possibly be elevated and a node could possibly be added to signify the "Tennis Membership" entity, which might then be linked to a number of different nodes.

Graphs of information particular to the consumer

It jogs my memory of the personalised search, however tells us that it isn’t restricted to the historical past of our analysis. It contains knowledge from sources reminiscent of e-mails that we might ship or obtain, or publications that we might submit on social media. This information graph might comprise details about the social hyperlinks now we have, but it surely additionally comprises data details about these hyperlinks. The patent tells us that personally identifiable data (together with location data) can even be protected.

And this tells us that the user-specific data graph data could possibly be put collectively to create a common data graph, which signifies that Google is constructing data graphs to reply particular questions and questions. for particular customers probably assembly to keep away from the boundaries of a graph of information primarily based on human-edited sources reminiscent of Wikipedia.

Share it’s love!