IDENTIFYING PURPOSE-BASED ORIGIN-DESTINATION USING CALL DETAILED RECORDS

• US 20130304363A1
• Filed: 05/08/2012
• Published: 11/14/2013
• Est. Priority Date: 05/08/2012
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Dai, Jing, Li, Ming, Murali, Pavankumar, Naphade, Milind R., Sahu, Sambit
• Granted Patent: US 9,042,908 B2
• Time in Patent Office: Days
• US Class Current: 701/117
• CPC Class Codes: G08G 1/012 from other sources than vehicle or roadside beacons, e.g. mobile networks
• CPC Class Codes: H04M 15/58 based on statistics of usage or network monitoring

Abstract

Access is obtained to call data records (or other data) for a plurality of mobile telephony users. The call data records (or other data) include location-time sequences. For each of the mobile telephony users, corresponding ones of the location-time sequences are segmented into trips; a home-work-school classification model is used to determine a home location and a work-school location, based on the trips; and a trip purpose classification model is used to assign each of the trips a purpose label, based at least on comparing at least one of a start location and an end location for each one of the trips to at least one of the home location and the work-school location determined using the home-work-school classification model. The trips are aggregated into purpose based origin-destination categories based at least on the purpose labels, time ranges of the location-time sequences, and the start and end locations.

Claims

1. A method comprising:

obtaining access to at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones for a plurality of mobile telephony users, said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones comprising location-time sequences;

for each of said mobile telephony users, segmenting corresponding ones of said location-time sequences into trips;

for each of said mobile telephony users, using a home-work-school classification model to determine a home location and a work-school location, based on said trips;

for each of said mobile telephony users, using a trip purpose classification model to assign each of said trips a purpose label, based at least on comparing at least one of a start location and an end location for each one of said trips to at least one of said home location and said work-school location determined using said home-work-school classification model; and

aggregating said trips into purpose based origin-destination categories based at least on said purpose labels, time ranges of said location-time sequences, and said start and end locations.

2. The method of claim 1, further comprising revising those of said trip purpose labels having a low confidence score, wherein said aggregating is based on said revised labels.

3. The method of claim 2, wherein, in said obtaining step, said location-time sequences are based at least on triangulated locations.

4. The method of claim 2, wherein, in said obtaining step, said location-time sequences are based at least on nearest cell tower locations.

5. The method of claim 2, wherein said segmenting step comprises segmenting based on time duration of a mobile device of a given one of said mobile telephony users in a given area, and wherein data for each of said trips is stored as start time, end time, start location, end duration, and stay duration.

6. The method of claim 2, wherein said using of said home-work-school classification model comprises using a rule-based home-work-school classification model based on prior knowledge.

7. The method of claim 2, wherein said using of said home-work-school classification model comprises training said home-work-school classification model based on labeled data from volunteers.

8. The method of claim 2, wherein said using of said home-work-school classification model comprises using a decision tree home-work-school classification model.

9. The method of claim 2, wherein said using of said home-work-school classification model comprises using a support vector machine (SVM) home-work-school classification model.

10. The method of claim 2, wherein said using of said home-work-school classification model comprises using an artificial neural network home-work-school classification model.

11. The method of claim 2, wherein said using of said trip purpose classification model comprises using a rule-based trip purpose classification model based on at least one of prior knowledge and points of interest.

12. The method of claim 2, wherein said using of said trip purpose classification model comprises training said trip purpose classification model based on labeled trips.

13. The method of claim 2, wherein said using of said trip purpose classification model comprises using a decision tree trip purpose classification model.

14. The method of claim 2, wherein said using of said trip purpose classification model comprises using a support vector machine (SVM) trip purpose classification model.

15. The method of claim 2, wherein said using of said trip purpose classification model comprises using an artificial neural network trip purpose classification model.

16. The method of claim 2, wherein:

said segmenting step comprises segmenting based on time duration of a mobile device of a given one of said mobile telephony users in a given area;

data for each of said trips is stored as start time, end time, start location, end location, and stay duration; and

said trip purpose classification model assigns each of said trips a purpose label further based on based at least on start time, end time, stay duration, and order of a given one of said trips during a day.

17. The method of claim 2, wherein said revising of said trip purpose labels comprises iteratively revising said trip purpose labels with a refined model based on associations among said trip purposes.

18. The method of claim 17, wherein said revising of said trip purpose labels comprises identifying for each of said trips additional attributes of previous purpose and next purpose.

19. The method of claim 2, further comprising at least one of:

routing at least one bus based on said purpose based origin-destination categories; and

plowing at least one road based on said purpose based origin-destination categories.

20. The method of claim 2, further comprising providing a system, wherein the system comprises distinct software modules, each of the distinct software modules being embodied in a non-transitory manner on a computer-readable storage medium, and wherein the distinct software modules comprise a database interface module, a segmentation module, a location classification engine module, a trip purpose engine module, and an aggregation module;

wherein;

said obtaining of said access to said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones is carried out by said database interface module executing on at least one hardware processor;

said segmenting of said corresponding ones of said location-time sequences into trips is carried out by said segmentation module executing on said at least one hardware processor;

said using of said home-work-school classification model to determine said home location and said work-school location is carried out by said location classification engine module executing on said at least one hardware processor;

said using of said trip purpose classification model is carried out by said trip purpose engine module executing on said at least one hardware processor;

said revising of said trip purpose labels is carried out by said trip purpose engine executing on said at least one hardware processor; and

said aggregating of said trips is carried out by said aggregation module executing on said at least one hardware processor.

21. The method of claim 1, further comprising, for each of said mobile telephony users, classifying a mode of movement based on trip start, trip end, trip duration, location of origin, location of destination, type of origin, type of destination, and information from other sources comprising at least one of available traffic information and meta data about telephone infrastructure collecting events.

22. A computer program product comprising a computer readable storage medium having computer readable program code embodied therewith in a non-transitory manner, said computer readable program code comprising:

computer readable program code configured to obtain access to at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones for a plurality of mobile telephony users, said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones comprising location-time sequences;

computer readable program code configured to, for each of said mobile telephony users, segment corresponding ones of said location-time sequences into trips;

computer readable program code configured to, for each of said mobile telephony users, use a home-work-school classification model to determine a home location and a work-school location, based on said trips;

computer readable program code configured to, for each of said mobile telephony users, use a trip purpose classification model to assign each of said trips a purpose label, based at least on comparing at least one of a start location and an end location for each one of said trips to at least one of said home location and said work-school location determined using said home-work-school classification model; and

computer readable program code configured to aggregate said trips into purpose based origin-destination categories based at least on said purpose labels, time ranges of said location-time sequences, and said start and end locations.

23. An apparatus comprising:

a memory; and

at least one processor, coupled to said memory, and operative to;

obtain access to at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones for a plurality of mobile telephony users, said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones comprising location-time sequences;

for each of said mobile telephony users, segment corresponding ones of said location-time sequences into trips;

for each of said mobile telephony users, use a home-work-school classification model to determine a home location and a work-school location, based on said trips;

for each of said mobile telephony users, use a trip purpose classification model to assign each of said trips a purpose label, based at least on comparing at least one of a start location and an end location for each one of said trips to at least one of said home location and said work-school location determined using said home-work-school classification model; and

aggregate said trips into purpose based origin-destination categories based at least on said purpose labels, time ranges of said location-time sequences, and said start and end locations.

24. The apparatus of claim 23, further comprising a plurality of distinct software modules, each of the distinct software modules being embodied in a non-transitory manner on a computer-readable storage medium, and wherein the distinct software modules comprise a database interface module, a segmentation module, a location classification engine module, a trip purpose engine module, and an aggregation module;

wherein;

said at least one processor is operative to obtain said access to said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones by executing said database interface module;

said at least one processor is operative to segment said corresponding ones of said location-time sequences into trips by executing said segmentation module;

said at least one processor is operative to use said home-work-school classification model to determine said home location and said work-school location by executing said location classification engine module;

said at least one processor is operative to use said trip purpose classification model by executing said trip purpose engine module;

said at least one processor is operative to revise said trip purpose labels by executing said trip purpose engine module; and

said at least one processor is operative to aggregate said trips by executing said aggregation module.

25. An apparatus comprising:

means for obtaining access to at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones for a plurality of mobile telephony users, said at least one of call data records, location update events, and control events between mobile telephone infrastructure and mobile phones comprising location-time sequences;

means for, for each of said mobile telephony users, segmenting corresponding ones of said location-time sequences into trips;

means for, for each of said mobile telephony users, using a home-work-school classification model to determine a home location and a work-school location, based on said trips;

means for, for each of said mobile telephony users, using a trip purpose classification model to assign each of said trips a purpose label, based at least on comparing at least one of a start location and an end location for each one of said trips to at least one of said home location and said work-school location determined using said home-work-school classification model; and

means for aggregating said trips into purpose based origin-destination categories based at least on said purpose labels, time ranges of said location-time sequences, and said start and end locations.