# US 8,543,533 B2 · Inferring influence and authority > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 8,543,533 B2 - Document ID: US8543533B2 - Status: Granted - Kind: B2 - Application number: 12/960,106 - Filed: 2010-12-03 - Published: 2012-06-07 - Granted: 2013-09-24 - Earliest claimed: 2010-12-03 - Assignees: International Business Machines Corporation - Applicants: Aurelie C. Lozano; Vikas Sindhwani - Inventors: Aurelie C. Lozano; Vikas Sindhwani - Web page: https://patentgrep.com/patent/US8543533B2 ## CPC - G06Q 30/02: Marketing; Price estimation or determination; Fundraising - G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR ## Abstract A computer model finds an originating community member or members, for instance amongst bloggers, scientific researchers, or phenomena in phenomenological systems. In one embodiment, non-explicit causal relationships may be inferred from comparing blogs as a whole. Influence relationships are derived from a weighted, directed graph output and sources of influence are ranked. This implementation is useful for a variety of applications. ## Claims ### Claim 1 A computer method comprising carrying out operations on at least one data processing device, the operations comprising: interfacing with a plurality of records of content originating from human community members; inferring at least one non-explicit causal relationship among at least two sets of records from the plurality of records, based on temporal sequence of appearance of content, the inferring including: for the at least two sets of records, calculating an adjacency matrix by using a multivariate granger causal modeling method which uses a matrix norm of a matrix whose elements include regression coefficients corresponding to variables which represent the at least two sets of records; and identifying, by using the calculated adjacency matrix, influence relationships among at least two originating community members, each community member associated with a set of records from the at least two sets of records. ### Claim 2 (depends on 1) The method of claim 1, wherein inferring comprises: considering for each set time dependent changes in a group of words belonging to that set; and determining whether or not the time dependent changes of any set as a whole are predictive of the time dependent changes of any other set as a whole. ### Claim 3 (depends on 2) The method of claim 2, wherein determining takes into account the contents of the other set as well as of those compared with that other set. ### Claim 4 (depends on 2) The method of claim 2, wherein each group comprises all the words belonging to the respective set. ### Claim 5 (depends on 2) The method of claim 2, wherein inferring comprises performing multivariate regression jointly for the sets of all the members with respect to past values of the sets of all the members. ### Claim 6 (depends on 2) The method of claim 2, wherein inferring comprises looping over each member, and for each member performing multivariate regression for the set corresponding to that member with respect to past values of the sets of all the members. ### Claim 7 (depends on 1) The method of claim 1, wherein each set comprises a set of words created by a person; persons are identified as influencers if the past values taken by their set of words are predictive of the present values of the set of words for other persons. ### Claim 8 (depends on 1) The method of claim 1, comprising outputting an embodiment of a weighted causal graph that encodes influence relationships between sources associated with content. ### Claim 9 (depends on 8) The method of claim 8, comprising, responsive to the graph, deriving a graph based ranking based on social influence measures. ### Claim 10 (depends on 9) The method of claim 9, wherein the influence measures are GrangerRanks. ### Claim 11 (depends on 1) The method of claim 1, wherein the relationship is Granger Causality. ### Claim 12 (depends on 1) The method of claim 1, wherein inferring comprises multivariate regression with variable group selection. ### Claim 13 (depends on 1) The method of claim 1, wherein inferring comprises multivariate regression with simultaneous variable group selection. ### Claim 14 (depends on 1) The method of claim 1, wherein inferring comprises Multivariate Group Orthogonal Matching Pursuit. ### Claim 15 (depends on 1) The method of claim 1, wherein the relationship is Judea Pearl causality. ### Claim 16 A computer program product for identifying influence, the computer program product comprising: a storage medium excluding only a propagating signal, the storage medium readable by a processing circuit and storing instructions to be run by the processing circuit for performing a method comprising: interfacing with a plurality of records of content originating from human community members; inferring at least one non-explicit causal relationship among at least two sets of records from the plurality of records, based on temporal sequence of appearance of content, the inferring including: for the at least two sets of records, calculating an adjacency matrix by using a multivariate granger causal modeling method which uses a matrix norm of a matrix whose elements include regression coefficients corresponding to variables which represent the at least two sets of records; and identifying, by using the calculated adjacency matrix, influence relationships among at least two originating community members, each community member associated with a set of records from the at least two sets of records. ### Claim 17 (depends on 16) The program product of claim 16, wherein inferring comprises: performing multivariate regression jointly for the sets of all the members with respect to past values of the sets of all the members, which includes: considering for each set the time dependent changes in all words from the group of words pertaining to that set as a whole; determining whether or not the time dependent changes of any set are predictive of the time dependent changes of any other set as a whole; each set comprises a set of words created by a person; and persons are identified as influencers if their time dependent changes of their sets are predictive of the time dependent changes of other sets. ### Claim 18 (depends on 17) The program product of claim 17, comprising outputting an embodiment of a weighted causal graph that encodes influence relationships between sources associated with content; and responsive to the graph, deriving influence measures. ### Claim 19 (depends on 17) The program product of claim 17, wherein inferring comprises multivariate regression with group variable selection. ### Claim 20 A system comprising: at least one medium for storing data; at least one interface for gathering data and supplying outputs; at least one processing device for carrying out operations, the operations comprising: interfacing with a plurality of records of content originating from human community members; inferring at least one non-explicit causal relationship among at least two sets of records from the plurality of records, based on temporal sequence of appearance of content, the inferring including: for the at least two sets of records, calculating an adjacency matrix by using a multivariate granger causal modeling method which uses a matrix norm of a matrix whose elements include regression coefficients corresponding to variables which represent the at least two sets of records; and identifying, by using the calculated adjacency matrix, influence relationships among at least two originating community members, each community member associated with a set of records from the at least two sets of records. ### Claim 21 (depends on 20) The system of claim 20, wherein the operations comprise outputting an embodiment of a weighted causal graph that encodes influence relationships between sources associated with content; and responsive to the graph, deriving influence measures. ### Claim 22 (depends on 21) The system of claim 21, wherein inferring comprises multivariate regression with group variable selection and joint consideration of all community members. ### Claim 23 (depends on 1) The method of claim 1, wherein the identifying includes: for each community member, computing a sum of entries in the adjacency matrix that departs from the each community member; ranking the community members according to the sum by which a community member with a larger sum has a higher rank.