Systems and methods for dynamic report generation based on automatic modeling of complex data structures

• US 10,445,152 B1
• Filed: 12/18/2015
• Issued: 10/15/2019
• Est. Priority Date: 12/19/2014
• Status: Active Grant

• Current Assignee: Experian Information Solutions Incorporated (Experian plc)
• Original Assignee: Experian Information Solutions Incorporated (Experian plc)
• Inventors: Zhang, Lingyun, Carpenter, Mason L., Bonin, Gregor, Xiong, Shanji, DiChiara, Christer, Zaleta, David, Tao, Yaqi
• Primary Examiner(s): Vincent, David R
• Application Number:
  US14/975,440
• Time in Patent Office: 1,397 Days
• Field of Search: 706 15, 706 45
• CPC Class Codes: G06F 16/1744 using compression, e.g. sparse files
• CPC Class Codes: G06F 16/23 Updating
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 9/542 Event management; Broadcasting; Multicasting; Notifications
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 20/38 Payment protocols; Details thereof
• CPC Class Codes: G06Q 20/4016 involving fraud or risk level assessment in transaction processing
• CPC Class Codes: G06Q 30/00 Commerce
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0204 Market segmentation

Abstract

Various systems and methods are disclosed for accessing and traversing disparate, complex, and multi-dimensional data structures to dynamically and interactively generate reports based on automated modeling of complex and non-uniformly formatted data. Automated analysis of probabilistic functions and temporal-based data records enable non-technical users to quickly and dynamically act on time-sensitive information. In response to various user inputs, the system automatically accesses and traverses complex data structures (including, for example, frequency distribution models) calculates complex data based on the traversals, displays the calculated complex data to the user, and enters the calculated complex data into the reports.

Claims

1. A computing system operable to access one or more electronic data sources in response to periodic automated inquiries in order to automatically calculate data for inclusion into a report, the computing system comprising:

a non-transitory storage device configured to store a plurality of event records associated with respective users, each of the event records indicating an event associated with a respective user; and

a physical processor that is in communication with the non-transitory storage device and that is configured to;

access the plurality of event records associated with respective users; and

for each individual event record of at least a subset of the event records, assign a category to the individual event record, the category selected from a plurality of predetermined categories;

generate a user profile for a particular user, wherein the user profile comprises categorized event records associated with the particular user during a set time period;

generate a user event frequency distribution model based on at least some of the categorized event records in the user profile of the particular user of a particular category, wherein the user event frequency distribution model predicts a likelihood that the particular user will engage in a future event in the particular category within a specified period of time;

access the generated user profile for the particular user and the user event frequency distribution model;

determine a gap for the particular user, the gap indicating a time period since a most recent event associated with the particular category by the particular user occurred;

determine a gap limit associated with the particular user, the gap limit indicating a period of time by which the particular user is expected to engage in the future event with the predicted likelihood based on the user event frequency distribution model;

compare the determined gap to the gap limit;

in response to determining that the gap is greater than the gap limit, trigger generation of an event change alert; and

transmit, to a client system, the generated event change alert indicating that the particular user has changed event behavior in the particular category.

2. The computing system of claim 1, wherein the physical processor is further configured to:

determine, in response to determining that the gap is greater than the gap limit, if a filter condition exists; and

in response to determining that a filter condition does not exist, trigger the generation of the event change alert.

3. The computing system of claim 1, wherein the gap limit is a period of time in which the particular user is expected to make a next event within a ninety-five percent (95%) probability, based on the user event frequency distribution model for the particular user.

4. The computing system of claim 1, wherein the physical processor is further configured to:

generate a category baseline event frequency distribution model for a particular category, the category baseline event frequency distribution model indicating a likelihood of an event in the particular category by a generic user based on a set of the accessed plurality of event records that are assigned to the particular category; and

update the category baseline event frequency distribution model for the particular category based on a set of the categorized event records of the particular user and associated with the particular category to generate a category-specific user event frequency distribution model.

5. The computing system of claim 4, wherein the physical processor is further configured to:

periodically access event data sources to determine whether there is an additional event record associated with the particular user and associated with the particular category; and

in response to determining that there is an additional event record associated with the particular user and associated with the particular category, update the category-specific user event frequency distribution model based on the determined additional event record.

6. The computing system of claim 4, wherein the physical processor is further configured to:

determine a second gap, indicating a time period since the most recent event by the particular user associated with the particular category occurred;

determine a second gap limit indicating a second expected period of time between events associated with the particular user and associated with the particular category;

compare the second gap to the second gap limit;

in response to determining that the second gap is greater than the second gap limit, trigger generation of a category-specific event change alert; and

transmit, to a client system, the category-specific event change alert.

7. The computing system of claim 1, wherein the physical processor is further configured to:

generate an event frequency distribution model for the additional event record associated with the particular user; and

calculate a weighted sum of the event distribution for the additional event record and the user event frequency distribution model to generate the updated user event frequency distribution model.

8. The computing system of claim 7, wherein the user event frequency distribution model for the additional event record associated with the particular user comprises a distribution having a one hundred percent (100%) probability of occurring within a time period between the most recent event by the particular user and a time of an event associated with the additional event record associated with the particular user.

9. The computing system of claim 7, wherein the user event frequency distribution model for the additional event record associated with the particular user comprises a distribution centered on a time period between the most recent event by the particular user and a time of an event associated with the additional event record.

10. The computing system of claim 1, wherein the physical processor is further configured to generate the event change alert comprising an identification of an event category associated with the event change alert, a number of days since the most recent event by the particular user occurred, and a number of events the particular user has made within a preceding two months.

11. The computing system of claim 1, wherein the physical processor is further configured to:

in response to determining that a filter condition exists, determine whether the filter condition is met; and

in response to determining that a filter condition is not met, generate an event change alert indicating that the gap is greater than the gap limit; and

transmit, to a client system, the event change alert, the event change alert including an identification of an event category associated with the event change alert, a number of days since a last event by the particular user occurred, and a number of events the particular user has made within a preceding two months.

12. The computing system of claim 1, further comprising a card reader in communication with the physical processor, the card reader including:

a payment information detector configured to detect payment information for an event of a user;

a targeted content generator configured to;

receive event data during the event of the user; and

identify content stored by the card reader using a comparison between a content selection rule and the event data, said content for presentation via the card reader; and

a display configured to present the content to the user.

13. A method of automatically generating a transaction frequency change alert, the method comprising:

accessing, from a transaction data store, a plurality of transaction records associated with respective users, the transaction records including attributes of a transaction made by the associated respective user;

accessing, from a categorized transaction data store, a transaction categories data structure including a plurality of transaction categories and, for each transaction category, attribute criteria usable to identify transactions associated with respective transaction categories;

for each of the accessed plurality of transaction records;

identifying one or more of the attributes of the transaction record;

comparing the identified one or more attributes of the transaction record to the attribute criteria of respective transaction categories to identify a transaction category among the plurality of transaction categories that matches the one or more attributes of the transaction record;

categorizing the accessed transaction record with the identified transaction category;

storing, in the categorized transaction data store, a plurality of categorized transaction records;

accessing, from the categorized transaction data store, the plurality of categorized transaction records;

determining, for each user and for each pair of timewise consecutive transactions of the user based on the accessed plurality of categorized transaction records, a time between transactions;

identifying, from the accessed plurality of categorized transaction records, a first plurality of categorized transaction records associated with a first user;

generate a first user transaction frequency distribution model based on at least some of the categorized transaction records associated with the first user of a particular category, wherein the first user transaction frequency distribution model predicts a likelihood that the first user will engage in a future transaction in the particular category within a specified period of time;

accessing the first user transaction frequency distribution model;

determining a time duration since a most recent transaction by the first user occurred associated with the particular category and comparing the determined time duration to a threshold period of time associated with the first user, the threshold period of time indicating a period of time by which the first user is expected to engage in the future transaction with the predicted likelihood;

in response to determining that the time duration is greater than the threshold period of time, triggering generation of a spend change alert; and

transmitting, to a client system, the generated spend change alert.

14. The method of claim 13, wherein determining a time duration since a last transaction by the first user occurred and comparing the determined time duration to a threshold period of time comprises using a threshold period of time based on the first user transaction frequency distribution model.

15. The method of claim 13, further comprising:

generating a baseline transaction frequency distribution model for a particular category, the baseline transaction frequency distribution model indicating a likelihood of a transaction in the particular category by a generic user, the model based on a set of the accessed plurality of categorized transaction records that are categorized to the particular category;

identifying, from the first plurality of categorized transaction records associated with the first user, a second plurality of categorized transaction records that are categorized to the particular category; and

updating the baseline transaction frequency distribution model for the particular category based on the second plurality of categorized transaction records that are categorized to the particular category to generate a first user category-specific transaction frequency distribution model.

16. The method of claim 15, further comprising:

determining a second time duration since the most recent transaction associated with the category by the first user occurred and a present time and comparing the second time duration to a category-specific threshold period of time indicating a second expected period of time between transactions by the first user in the particular category; and

in response to determining that the second time duration is greater than the category-specific threshold period of time, generating a spend change alert indicating that the second time duration is greater than the category-specific threshold period of time.

17. The method of claim 13, wherein updating the first user transaction frequency distribution model based on the additional categorized transaction record comprises:

generating a transaction frequency distribution model for the additional categorized transaction record associated with the first user which has not been analyzed; and

calculating a weighted sum of the generated transaction frequency distribution model for the additional transaction record and the first user transaction frequency distribution model to generate the updated first user transaction frequency distribution model.

18. The method of claim 17, wherein generating a transaction frequency distribution model for the additional categorized transaction record associated with the first user which has not been analyzed comprises generating a distribution having a one hundred percent (100%) probability of occurring within a time period between the most recent transaction by the first user and a time of a transaction associated with the additional transaction record associated with the first user.

19. The method of claim 17, wherein generating a transaction frequency distribution model for the additional categorized transaction record associated with the first user which has not been analyzed by the system comprises generating a distribution centered on a time period between the most recent transaction by the first user and a time of a transaction associated with the additional transaction record associated with the first user.