Managing information about entities using observations

• US 9,940,381 B1
• Filed: 02/06/2015
• Issued: 04/10/2018
• Est. Priority Date: 07/12/2011
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google LLC (Alphabet Inc.)
• Inventors: Janos, Joseph, Strohm, Alan C., Harb, Boulos, Stern, Steven M., Sahuguet, Arnaud, de Alvarenga Oliveira, Ademir
• Primary Examiner(s): DWIVEDI, MAHESH H
• Application Number:
  US14/616,321
• Time in Patent Office: 1,159 Days
• Field of Search: 707705, 707737
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/958 Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for managing entities using observations. In one aspect, a method includes receiving data identifying an entity; generating a user interface document that, when rendered by a user device, presents a plurality of attribute values to a user and allows the user to modify one or more of the plurality of attribute values; and storing an observation in a data store, the observation including a user-modified value of one of the plurality of attribute values and a context including one or more of the presented attribute values.

Claims

1. A method, comprising:

obtaining a collection of clusters of immutable observations about entities, at least a plurality of the clusters each;

corresponding to a respective entity,

identifying immutable observations determined to describe the respective entity, and

having summary attribute-value pairs that summarize the identified observations and collectively describe an inferred current state of the respective entity;

receiving a new observation about a given entity;

selecting, with one or more processors, a cluster among the collection of clusters based on correspondence of the selected cluster to the given entity;

summarizing, with one or more processors, the selected cluster by updating at least some of the summary attribute-value pairs of the selected cluster based on both the new observation and the observations identified by the selected cluster;

storing the updated attribute-value pairs in memory in association with the selected cluster; and

updating a user interface at a user device using the new observation.

2. The method of claim 1, wherein:

summarizing the selected cluster comprises comparing at least part of the new observation to observations identified by the selected cluster, wherein at least some of the observations identified by the selected cluster are inconsistent with the summary attribute-value pairs of the selected cluster.

3. The method of claim 1, wherein:

the new observation comprises;

a plurality of context attribute-value pairs that are consistent with corresponding summary attribute-value pairs of the selected cluster; and

a payload value of a given attribute that is different from a corresponding summary attribute-value pair of the selected cluster;

selecting a cluster among the collection of clusters comprises selecting the cluster based on the plurality of context attribute-value pairs; and

updating at least some of the summary attribute-value pairs of the selected cluster comprises;

determining that the payload value is more accurate than a value of the corresponding summary attribute-value pair of the selected cluster; and

in response to the determination, updating the value of the corresponding summary attribute-value pair to reflect the payload value of the new observation.

4. The method of claim 1, comprising:

sending the user device instructions that cause the user device to display at least some of the summary attribute-value pairs of the given entity,

wherein the new observation is received from the user device and at least some of the identified observations comprise;

a payload containing one or more attribute-value pairs designated by the user as needing to be changed or added to the summary attribute value pairs of the given entity; and

a context containing one or more attribute-value pairs displayed to the user and not designated by the user as needing to be changed or added.

5. The method of claim 1, wherein:

at least some of the entities are businesses; and

at least some of the summary attribute value pairs, in the aggregate, describe the following;

a phone number of a business;

a geographic location of a business;

a name of a business; and

operating hours of a business.

6. The method of claim 1, wherein obtaining a collection of clusters comprises:

receiving feeds from data providers, each feed having a plurality of observations about entities, at least some of the observations being inconsistent with one another;

clustering the received observations from the feeds to form at least part of the collection of clusters, at least some of the resulting clusters identifying inconsistent observations about the respective corresponding entity; and

resolving the inconsistent observations in summary attribute-value pairs of the resulting clusters.

7. The method of claim 1, wherein

the new observation comprises a context;

and

selecting a cluster among the collection of clusters comprises matching the new observation to the selected cluster based on at least two of the following;

a comparison between an entity street address described by the context and a street address described by a summary attribute value pair of the selected cluster;

determining a geographic distance between an entity geolocation described by the context and a geolocation described by a summary attribute value pair of the selected cluster;

a comparison between an entity phone number described by the context and a phone number described by a summary attribute value pair of the selected cluster; and

determining an edit distance between an entity title described by the context and a title described by a summary attribute value pair of the selected cluster.

8. The method of claim 1, wherein:

the new observation includes a user-generated value of an attribute of the given entity and a context that includes the values of one or more other attributes of the given entity; and

selecting a cluster among the collection of clusters as corresponding to the given entity comprises;

composing at least part of a query based on one more of the values of attributes included in the context;

identifying candidate clusters responsive to the query based on the candidate clusters having summary attribute-value pairs that match the attribute values upon which the query is based;

calculating a respective score for each of the candidate clusters; and

associating the new observation with a candidate cluster selected based on the calculated scores.

9. The method of claim 1, wherein summarizing the selected cluster by updating at least some of the summary attribute-value pairs of the selected cluster comprises:

determining that a state of the given entity has changed; and

in response to the determination, updating a summary attribute-value pair of the cluster corresponding to the given entity based on a value conveyed by the new observation.

10. The method of claim 1, wherein summarizing the selected cluster by updating at least some of the summary attribute-value pairs of the selected cluster comprises:

determining that a value of the new observation is reliable.

11. The method of claim 1, wherein summarizing the selected cluster by updating at least some of the summary attribute-value pairs of the selected cluster comprises:

determining that a user submitting the new observation is trustworthy based on a trust score of the user.

12. The method of claim 1, comprising:

determining based on the new observation that two distinct entities are currently being represented by one cluster in the collection and, in response, re-clustering observations that have been previously matched to the one cluster to produce two clusters in the collection.

13. The method of claim 1, comprising:

determining based on the new observation that the given entity is currently being represented by two or more clusters in the collection and, in response, re-clustering observations that have been previously matched to the two or more clusters to produce a single consolidated cluster.

14. The method of claim 1, comprising:

determining based on the new observation that one or more observations were incorrectly matched to their current cluster, and in response, re-clustering at least some of the one or more observations into a different, existing cluster in the collection.

15. The method of claim 1, comprising:

ascertaining trustworthiness of a user submitting the new observation; and

updating other observations submitted by the user identified in the collection to indicate a changed trust score for the user.

16. The method of claim 1, comprising:

receiving, from a requestor, a request for information about an entity corresponding to a given cluster in the collection;

in response to the request, sending the requestor at least some of the summary attribute-value pairs of the given cluster.

17. The method of claim 1, wherein:

some clusters in the collection have a single observation;

the observations are immutable;

the collection is stored in a cluster index;

selecting a cluster among the collection of clusters comprises matching the new observation to the selected cluster based on all of the following;

a comparison between an entity street address described by the context and a street address described by a summary attribute value pair of the selected cluster;

determining a geographic distance between an entity geolocation described by the context and a geolocation described by a summary attribute value pair of the selected cluster;

a comparison between an entity phone number described by the context and a phone number described by a summary attribute value pair of the selected cluster; and

determining an edit distance between an entity title described by the context and a title described by a summary attribute value pair of the selected cluster;

the method comprises sending a user information about the given entity and modifying the sent information based on the new observation, wherein the new observation is received from the user after sending the information about the given entity.

18. A system, comprising:

one or more processors;

memory storing instructions that when executed by at least some of the one or more processors effectuates operations comprising;

obtaining a collection of clusters of immutable observations about entities, at least a plurality of the clusters each;

corresponding to a respective entity,

identifying immutable observations determined to describe the respective entity, and

having summary attribute-value pairs that summarize the identified observations and collectively describe an inferred current state of the respective entity;

receiving a new observation about a given entity;

selecting a cluster among the collection of clusters based on correspondence of the selected cluster to the given entity;

summarizing the selected cluster by updating at least some of the summary attribute-value pairs of the selected cluster based on both the new observation and the observations identified by the selected cluster;

storing the updated attribute-value pairs in association with the selected cluster; and

updating a user interface at a user device using the new observation.

19. The system of claim 18, wherein:

summarizing the selected cluster comprises comparing at least part of the new observation to observations identified by the selected cluster, wherein at least some of the observations identified by the selected cluster are inconsistent with the summary attribute-value pairs of the selected cluster.

20. The system of claim 18, wherein:

the new observation comprises;

a plurality of context attribute-value pairs that are consistent with corresponding summary attribute-value pairs of the selected cluster; and

a payload value of a given attribute that is different from a corresponding summary attribute-value pair of the selected cluster;

selecting a cluster among the collection of clusters comprises selecting the cluster based on the plurality of context attribute-value pairs; and

updating at least some of the summary attribute-value pairs of the selected cluster comprises;

determining that the payload value is more accurate than a value of the corresponding summary attribute-value pair of the selected cluster; and

in response to the determination, updating the value of the corresponding summary attribute-value pair to reflect the payload value of the new observation.