User demographics determination based on commonality in call graphs
View Patent ↗A method of operating a data processing system to infer demographic information for a user of a wireless communication device comprises processing a plurality of call detail records (CDRs) to generate a plurality of call graphs having different time slices based on time ranges when data in the CDRs was collected, wherein the call graphs comprise nodes that represent individual callers and edges between the nodes that represent bi-directional communication between the individual callers. The method further comprises identifying neighbors of the user that have a high likelihood of sharing a common demographic attribute with the user based on communication features and structural features among the neighbors on one of the call graphs. The method further comprises identifying a most similar neighbor among the neighbors of the user that has a highest likelihood of sharing the common demographic attribute with the user.
1. A method of operating a data processing system to infer demographic information for a user of a wireless communication device, the method comprising:
processing a plurality of call detail records (CDRs) to generate a plurality of call graphs having different time slices based on time ranges when data in the CDRs was collected, wherein the call graphs comprise nodes that represent individual callers and edges between the nodes that represent bi-directional communication between the individual callers;
identifying neighbors of the user that have a high likelihood of sharing a common demographic attribute with the user based on communication features and structural features among the neighbors on one of the call graphs, wherein the common demographic attribute comprises a similarity of a home location of the user and the neighbors; and
identifying a most similar neighbor among the neighbors of the user that has a highest likelihood of sharing the common demographic attribute with the user by at least identifying a neighbor that has a shortest average call duration with the user and a greatest number of common contacts shared with the user.
2. The method of claim 1 wherein the call graphs having the different time slices based on the time ranges when the data in the CDRs was collected comprise a weekday call graph, a nights and weekends call graph, and a complete call graph.
3. The method of claim 1 wherein the edges between the nodes include a number of calls between the nodes, an average call duration of the calls, a degree difference between the nodes that represents a difference between a total number of contacts with which each node communicates, and a number of common contacts between the nodes.
4. The method of claim 1 wherein the common demographic attribute further comprises a similarly of an age of the user and the neighbors.
5. The method of claim 4 wherein identifying the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarly of the age of the user comprises identifying a neighbor that has a greatest number of weekday calls with the user.
6. The method of claim 4 wherein identifying the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarly of the age of the user comprises identifying a neighbor that has a shortest average call duration with the user.
7. The method of claim 4 wherein identifying the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarly of the age of the user comprises identifying a neighbor that has a greatest number of common contacts shared with the user during weekdays.
8. The method of claim 1 wherein the common demographic attribute comprises a similarly of an income level of the user and the neighbors.
9. A computer apparatus to infer demographic information for a user of a wireless communication device, the apparatus comprising:
software instructions configured, when executed by a data processing system, to direct the data processing system to:
process a plurality of call detail records (CDRs) to generate a plurality of call graphs having different time slices based on time ranges when data in the CDRs was collected, wherein the call graphs comprise nodes that represent individual callers and edges between the nodes that represent bi-directional communication between the individual callers;
identify neighbors of the user that have a high likelihood of sharing a common demographic attribute with the user based on communication features and structural features among the neighbors on one of the call graphs, wherein the common demographic attribute comprises a similarly of an age of the user and the neighbors; and
identify a most similar neighbor among the neighbors of the user that has a highest likelihood of sharing the common demographic attribute with the user by at least identifying a neighbor that has a shortest average call duration with the user; and
at least one non-transitory computer-readable storage medium storing the software instructions.
10. The computer apparatus of claim 9 wherein the call graphs having the different time slices based on the time ranges when the data in the CDRs was collected comprise a weekday call graph, a nights and weekends call graph, and a complete call graph.
11. The computer apparatus of claim 9 wherein the edges between the nodes include a number of calls between the nodes, an average call duration of the calls, a degree difference between the nodes that represents a difference between a total number of contacts with which each node communicates, and a number of common contacts between the nodes.
12. The computer apparatus of claim 9 wherein the common demographic attribute further comprises a similarity of a home location of the user and the neighbors.
13. The computer apparatus of claim 12 wherein the software instructions configured to direct the data processing system to identify the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarity of the home location with the user comprises the software instructions configured to direct the data processing system to identify a neighbor that has a shortest average call duration with the user and a greatest number of common contacts shared with the user.
14. The computer apparatus of claim 9 wherein the software instructions configured to direct the data processing system to identify the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarly of the age of the user comprises the software instructions configured to direct the data processing system to identify a neighbor that has a greatest number of weekday calls with the user.
15. The computer apparatus of claim 9 wherein the software instructions configured to direct the data processing system to identify the most similar neighbor among the neighbors of the user that has the highest likelihood of sharing the similarly of the age of the user comprises the software instructions configured to direct the data processing system to identify a neighbor that has a greatest number of common contacts shared with the user during weekdays.
16. The computer apparatus of claim 9 wherein the common demographic attribute comprises a similarly of an income level of the user and the neighbors.