Analysis, inference, and visualization of social networks

• US 8,862,622 B2
• Filed: 12/10/2008
• Issued: 10/14/2014
• Est. Priority Date: 12/10/2007
• Status: Active Grant

• Current Assignee: Highground Technologies Corporation
• Original Assignee: Sprylogics International Corp.
• Inventors: Zivkovic, Aleksandar, Shachar, Avichai
• Primary Examiner(s): Girma, Anteneh
• Assistant Examiner(s): MUELLER, KURT A
• Application Number:
  US12/332,046
• Publication Number:
  US 20090164431A1
• Time in Patent Office: 2,134 Days
• Field of Search: 707/732, 707/780
• US Class Current: 707/780
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: G06F 7/24 Sorting, i.e. extracting data from one or more carriers, rearranging the data in numerical or other ordered sequence, and rerecording the sorted data on the original carrier or on a different carrier or set of carriers sorting methods in generalG06F7/36 takes precedence
• CPC Class Codes: G06Q 10/10 Office automation; Time management

Abstract

A method and system for automated generation of social networks. A graphical user interface receives a user query for an entity of interest, and outputs a graphical network showing entities and associations related to the entity of interest. A search engine interface transmits the query to a search engine, and receives references to documents. A named entity extractor downloads a selection of the documents, and generates a list of named entities referenced in the downloaded documents. A network inference module receives each list of named entities, and generates associations between the named entities in each list. An entity matcher operates on the associations to consolidate them in instances wherein differently named entities are determined to be the same named entity, and provides a consolidated list of named entities and associations to the user interface for display as a graphical network.

Claims

1. A computer implemented system for inferring social networks, the system comprising:

a computer having a hardware microprocessor, the hardware microprocessor being configured to execute;

a graphical user interface that receives, from a user, a query relating to an entity of interest other than the user, and that responsively provides, to the user, a graphical network of entities and associations related to the entity of interest;

a search engine interface that transmits, over the communication network to a search engine, the query received by the graphical user interface, and that responsively receives, over the communication network from the search engine, search results that refer to a plurality of digitally encoded documents related to the entity of interest;

a named entity extractor that downloads, over the communication network, a selection of the referenced digitally encoded documents, and that extracts named entities from the downloaded documents;

a network inference module that constructs an inferred network of associations between the extracted named entities from lexical analysis of text of the downloaded documents; and

an entity matcher that operates on the associations generated by the network inference module so as to disambiguate differently named entities when the differently named entities are determined to be the same entity, and that provides a resulting list of disambiguated named entities and the network of associations between the disambiguated named entities to the graphical user interface for display as the graphical network;

wherein the network inference module operates by;

breaking up the text of the downloaded documents into a plurality of sentences;

within each sentence in the plurality of sentences, identifying extracted named entities and, when at least two such named entities are within the sentence, identifying a textual association between each pair of the named entities; and

storing the extracted named entities and associations as the inferred network.

2. A system according to claim 1, further comprising a risk evaluator, coupled to the entity matcher, that uses a digital process to evaluate a risk factor for the entity of interest, the risk factor comprising a weighted sum of a plurality of risk values, each risk value in the plurality of risk values being associated with either (i) the entity of interest or (ii) an associated, extracted named entity.

3. A system according to claim 1, wherein the entity of interest is selected from the group consisting of a person, a company, a location, an event, a date and a phone number.

4. A system according to claim 1, wherein the associations between the entities are selected from the group consisting of a family relationship, a business partnership, ownership, a legal relationship and a financial relationship.

5. A system according to claim 1, wherein the search engine is an Internet search engine, and wherein the plurality of documents is a plurality of web documents.

6. A method for a computer implemented system for inferring social networks, the method comprising:

receiving, from a user via a graphical user interface, a query relating to an entity of interest other than the user;

transmitting the query over a communication network to a search engine;

responsively receiving, over the communication network from the search engine, search results that refer to a plurality of digitally encoded documents related to the entity of interest;

downloading, over the communication network, a selection of the referenced digitally encoded documents;

extracting named entities from the downloaded selection of documents;

constructing an inferred network of associations between the extracted named entities from lexical analysis of the text of the downloaded selection of documents;

disambiguating differently named entities when the differently named entities are determined to be the same entity; and

providing, to the graphical user interface for display as a graphical network, the disambiguated named entities and the network of associations between the disambiguated named entities;

wherein constructing the inferred network includes;

breaking up the text of the downloaded selection of documents into a plurality of sentences;

within each sentence in the plurality of sentences, identifying extracted named entities and, when at least two such named entities are within the sentence, identifying a textual association between each pair of the named entities; and

storing the extracted named entities and associations as the inferred network.

7. A method according to claim 6, further comprising computing a risk factor for the entity of interest based on the named entities and the associations between the named entities, the risk factor comprising a weighted sum of a plurality of risk values, each risk value in the plurality of risk values being associated with either (i) the entity of interest or (ii) an associated, extracted named entity.

8. A method according to claim 6, wherein the entity of interest is selected from the group consisting of a person, a company, a location, an event, a date and a phone number.

9. A method according to claim 6, wherein the associations between the entities are selected from the group consisting of a family relationship, a business partnership, ownership, a legal relationship and a financial relationship.

10. A method according to claim 6, wherein the search engine is an Internet search engine, and wherein the plurality of documents is a plurality of web documents.

11. A computer implemented method for computing a risk factor for an entity of interest, comprising:

retrieving a digitally encoded and stored social network of entities related to an entity of interest, and associations between such entities, wherein the social network is constructed from named entities extracted from digitally encoded documents, by lexical analysis of the text of the digitally encoded documents, the documents having been downloaded from a communication network in response to a query regarding the entity of interest by a user other than the entity of interest to a search engine over the communication network, the social network represented by a digitally encoded graph whose vertices represent entities and whose edges represent associations between entities; and

in a digital process, deriving a risk factor for the entity of interest, the risk factor comprising a weighted sum of contributions from a plurality of individual paths, each individual path comprising one or more associations that traverse the graph from the entity of interest to each of the entities related thereto;

wherein constructing the social network includes;

breaking up the text of the digitally encoded documents into a plurality of sentences;

within each sentence in the plurality of sentences, identifying extracted named entities and, when at least two such named entities are within the sentence, identifying a textual association between each pair of the named entities; and

storing the extracted named entities and associations as vertices and edges, respectively, of the social network.

12. A method according to claim 11 wherein deriving said weighted sum comprises summing contributions from individual paths comprising no more than a specified number of associations.

13. A method according to claim 11 wherein the contribution from an individual path depends on the number of associations that comprise the path.

14. A method according to claim 11 wherein the contribution from an individual path depends on a known intrinsic risk of the related entity at the end of the path.

15. A method according to claim 11 wherein the contribution from an individual path depends on the nature of each association comprising the path.

16. A computer implemented method for analyzing differently named entities, comprising:

downloading over a communication network two digitally encoded documents which each refer to differently named entities;

for each of the two documents, deriving social contexts of the differently named entities from lexical analysis of the text of the downloaded documents;

comparing the derived social contexts for overlap; and

determining that the differently named entities refer to the same entity based on the results of said comparing if the overlap is greater than a predetermined threshold;

wherein deriving the social contexts includes;

breaking up the text of the downloaded documents into a plurality of sentences;

within each sentence in the plurality of sentences, identifying extracted named entities and, when at least two such named entities are within the sentence, identifying a textual association between each pair of the named entities; and

storing the extracted named entities and associations.

17. A computer program product for inferring social networks, the computer program product comprising a tangible, non-transitory computer usable medium having computer readable program code thereon, the computer readable program code comprising:

program code for receiving, from a user via a graphical user interface, a query relating to an entity of interest other than the user;

program code for causing the query to be transmitted over a communication network to a search engine;

program code for responsively receiving, over the communication network from the search engine, search results that refer to a plurality of digitally encoded documents related to the entity of interest;

program code for causing a selection of the referenced digitally encoded documents to be downloaded over the communication network;

program code for extracting named entities from the downloaded selection of documents;

program code for constructing an inferred network of associations between the extracted named entities from lexical analysis of the text of the downloaded documents;

program code for disambiguating differently named entities when the differently named entities are determined to be the same entity; and

program code for providing, to the graphical user interface for display as a graphical network, the disambiguated named entities and the network of associations between the disambiguated named entities;

wherein the program code for constructing the inferred network includes;

program code for breaking up the text of the downloaded documents into a plurality of sentences;

program code for, within each sentence in the plurality of sentences, identifying extracted named entities and, when at least two such named entities are within the sentence, identifying textual association between each pair of the named entities; and

program code for storing the extracted named entities and associations as the inferred network.

18. A computer program product according to claim 17, further comprising program code for computing a risk factor for the entity of interest based on the named entities and the associations between the named entities, the risk factor comprising a weighted sum of a plurality of risk values, each risk value in the plurality of risk values being associated with either (i) the entity of interest or (ii) an associated, extracted named entity.