Determining key concepts in documents based on a universal concept graph

• US 10,255,282 B2
• Filed: 03/04/2016
• Issued: 04/09/2019
• Est. Priority Date: 03/04/2016
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Kenthapadi, Krishnaram, Borisyuk, Fedor Vladimirovich, Jain, Parul
• Primary Examiner(s): Phillips, III, Albert M
• Application Number:
  US15/061,296
• Publication Number:
  US 20170255621A1
• Time in Patent Office: 1,131 Days
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/9024 Graphs; Linked lists G06F16/9027 takes precedence
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/972 Access to data in other repository systems, e.g. legacy data or dynamic Web page generation
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: H04L 67/01 Protocols
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/306 User profiles
• CPC Class Codes: H04L 67/53 using third party service providers

Abstract

A machine may be configured to determining key concepts in documents. For example, the machine accesses a universal concept graph that includes a first set of nodes that represent concept phrases derived from internal documents associated with a social networking service (SNS) and external documents that are external to the SNS, and a first set of edges that connect a plurality of nodes of the first set of nodes. The machine accesses a content object associated with the SNS. The machine generates an induced concept graph associated with the content object based on an analysis of the content object and the universal concept graph. The machine identifies one or more key concept phrases in the content object based on applying one or more key concept selection algorithms to the induced concept graph. The machine stores the one or more key concept phrases in a record of a database.

Claims

1. A method comprising:

accessing a universal concept graph that includes a first set of nodes that represent concept phrases derived from one or more internal documents associated with a social networking service (SNS) and one or more external documents that are external to the SNS, and a first set of edges that connect a plurality of nodes of the first set of nodes;

accessing a content object associated with the SNS;

generating, using one or more hardware processors, an induced concept graph associated with the content object based on analysis of the content object and the universal concept graph, the induced graph including a second set of nodes that represent one or more concept phrases derived from the content object and a second set of edges that connect a plurality of nodes of the second set of nodes;

identifying one or more key concept phrases in the content object based on applying one or more key concept selection algorithms to the induced concept graph, a first key concept selection algorithm of the one or more key concept selection algorithms iteratively removing leaf nodes from the induced concept graph associated with the content object until a desired number of nodes representing key concept phrases are left, a leaf node being connected solely to one other node; and

storing the one or more key concept phrases in a record of a database, the record referencing the content object.

2. The method of claim 1, wherein the identifying of the one or more key concept phrases includes:

selecting the key concept phrases from the one or more concept phrases derived from the content object, the selecting being performed according to a further key concept selection algorithm of the one or more key concept selection algorithms;

the method further comprising;

generating one or more intermediate rankings of the key concept phrases of the content object according to the further key concept selection algorithm.

3. The method of claim 2, further comprising:

generating a combined ranking of the key concept phrases based on combining the one or more intermediate rankings of the key concept phrases of the content object according to a rank aggregation technique.

4. The method of claim 1, wherein the content object includes a field of a member profile of a member of the SNS, and wherein the one or more key concept phrases are identified in the field of the member profile.

5. The method of claim 4, wherein the field of the member profile is a first field of the member profile, and

wherein the identifying of the one or more key concept phrases in the content object includes identifying a set of key concept phrases in the member profile based on identifying one or more key concept phrases in the first field of the member profile and identifying one or more key concept phrases in a second field of the member profile.

6. The method of claim 1, further comprising:

causing a presentation of at least a portion of the induced concept graph in a user interface of a device associated with a user.

7. The method of claim 1, wherein the generating of the induced concept graph associated with the content object includes:

generating a set of tokens based on the content object;

mapping one or more tokens of the set of tokens to one or more nodes of the first set of nodes included in the universal concept graph; and

generating a candidate set of concept phrases for the content object based on the mapping of the one or more tokens of the set of tokens to the one or more nodes of the first set of nodes.

8. The method of claim 1, wherein the generating of the induced concept graph associated with the content object includes:

accessing a candidate set of concept phrases for the content object;

mapping a pair of concept phrases of the candidate set to a first pair of nodes in the universal concept graph;

identifying a first edge of the first set of edges that connects the first pair of nodes in the universal concept graph;

generating the second set of edges to be included in the induced concept graph, the generating being based on the identified first edge that connects the first pair of nodes in the universal concept graph, the second set of edges including a second edge to connect a second pair of nodes corresponding to the pair of concept phrases of the candidate set.

9. The method of claim 8, further comprising:

based on the candidate set of concept phrases for the content object and the second edge, generating the second set of nodes to be included in the induced concept graph, the second set of nodes including the second pair of nodes to be connected by the second edge in the induced concept graph associated with the content object; and

generating the induced concept graph based on the second set of nodes and the second set of edges, the induced concept graph including the second pair of nodes connected by the second edge.

10. The method of claim 8, further comprising:

identifying an edge weight value associated with the first edge of the first set of edges; and

associating the edge weight value with the second edge to connect the second pair of nodes.

11. A system comprising

one or more hardware processors;

and

a non-transitory machine-readable medium for storing instructions that, when executed by one or more hardware processors, cause the system to perform operations comprising;

accessing a universal concept graph that includes a first set of nodes that represent concept phrases derived from one or more internal documents associated with a social networking service (SNS) and one or more external documents that are external to the SNS, and a first set of edges that connect a plurality of nodes of the first set of nodes;

accessing a content object associated with the SNS;

generating an induced concept graph associated with the content object based on analysis of the content object and the universal concept graph, the induced graph including a second set of nodes that represent one or more concept phrases derived from the content object and a second set of edges that connect a plurality of nodes of the second set of nodes;

identifying one or more key concept phrases in the content object based on applying one or more key concept selection algorithms to the induced concept graph, a first key concept selection algorithm of the one or more key concept selection algorithms iteratively removing leaf nodes from the induced concept graph associated with the content object until a desired number of nodes representing key concept phrases are left, a leaf node being connected solely to one other node; and

storing the one or more key concept phrases in a record of a database, the record referencing the content object.

12. The system of claim 11, wherein the identifying of the one or more key concept phrases includes:

selecting the key concept phrases from the one or more concept phrases derived from the content object, the selecting being performed according to a further key concept selection algorithm of the one or more key concept selection algorithms,

wherein the operations further comprise;

generating one or more intermediate rankings of the key concept phrases of the content object according to the further key concept selection algorithm.

13. The system of claim 12, wherein the operations further comprise:

generating a combined ranking of the key concept phrases based on combining the one or more intermediate rankings of the key concept phrases of the content object according to a rank aggregation technique.

14. The system of claim 11, wherein the content object includes a field of a member profile of a member of the SNS, and wherein the one or more key concept phrases are identified in the field of the member profile.

15. The system of claim 14, wherein the field of the member profile is a first field of the member profile, and

wherein the identifying of the one or more key concept phrases in the content object includes identifying a set of key concept phrases in the member profile based on identifying one or more key concept phrases in the first field of the member profile and identifying one or more key concept phrases in a second field of the member profile.

16. The system of claim 11, wherein the generating of the induced concept graph associated with the content object includes:

generating a set of tokens based on the content object;

mapping one or more tokens of the set of tokens to one or more nodes of the first set of nodes included in the universal concept graph; and

generating a candidate set of concept phrases for the content object based on the mapping of the one or more tokens of the set of tokens to the one or more nodes of the first set of nodes.

17. The system of claim 11, wherein the generating of the induced concept graph associated with the content object includes:

accessing a candidate set of concept phrases for the content object;

mapping a pair of concept phrases of the candidate set to a first pair of nodes in the universal concept graph;

identifying a first edge of the first set of edges that connects the first pair of nodes in the universal concept graph;

generating the second set of edges to be included in the induced concept graph, the generating being based on the identified first edge that connects the first pair of nodes in the universal concept graph, the second set of edges including a second edge to connect a second pair of nodes corresponding to the pair of concept phrases of the candidate set.

18. The system of claim 17, wherein the operations further comprise:

based on the candidate set of concept phrases for the content object and the second edge, generating the second set of nodes to be included in the induced concept graph, the second set of nodes including the second pair of nodes to be connected by the second edge in the induced concept graph associated with the content object; and

generating the induced concept graph based on the second set of nodes and the second set of edges, the induced concept graph including the second pair of nodes connected by the second edge.

19. The system of claim 17, wherein the operations further comprise:

identifying an edge weight value associated with the first edge of the first set of edges; and

associating the edge weight value with the second edge that connects the second pair of nodes included in the induced concept graph.

20. A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more hardware processors of a machine, cause the machine to perform operations comprising:

accessing a universal concept graph that includes a first set of nodes that represent concept phrases derived from one or more internal documents associated with a social networking service (SNS) and one or more external documents that are external to the SNS, and a first set of edges that connect a plurality of nodes of the first set of nodes;

accessing a content object associated with the SNS;

generating an induced concept graph associated with the content object based on analysis of the content object and the universal concept graph, the induced graph including a second set of nodes that represent one or more concept phrases derived from the content object and a second set of edges that connect a plurality of nodes of the second set of nodes;

identifying one or more key concept phrases in the content object based on applying one or more key concept selection algorithms to the induced concept graph, a first key concept selection algorithm of the one or more key concept selection algorithms iteratively removing leaf nodes from the induced concept graph associated with the content object until a desired number of nodes representing key concept phrases are left, a leaf node being connected solely to one other node; and

storing the one or more key concept phrases in a record of a database, the record referencing the content object.