System and method for revealing correlations between data streams

• US 8,762,302 B1
• Filed: 02/20/2014
• Issued: 06/24/2014
• Est. Priority Date: 02/22/2013
• Status: Active Grant

• Current Assignee: Blazent Incorporated (ServiceNow Incorporated)
• Original Assignee: Bottlenose, Inc. (ServiceNow Incorporated)
• Inventors: Spivack, Nova, ter Heide, Dominiek, Mousavi, Seyed Ali
• Primary Examiner(s): CHANG, LI WU
• Application Number:
  US14/185,809
• Time in Patent Office: 124 Days
• US Class Current: 706/12
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 5/046 Forward inferencing; Production systems
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

The disclosed techniques can provide users with a tool having an integrated, user-friendly interface and having automated mechanisms which can reveal correlations between data streams to the users in a clear and easily understandable way, thereby enabling the users to easily digest the vast amount of information contained in activities within one or more network, to understand the correlations among the activities, to stay informed and responsive to current or new trends, and even to predict future trends. Among other benefits, the disclosed techniques are especially useful in the context of discovering impacts of social networking activities on other types of commercial activities.

Claims

1. A method for revealing correlations in data signals, the method comprising:

automatically importing a first data stream comprising a first time series of the data signals from conversations occurred within one or more social networks;

importing a second data stream comprising a second time series of the data signals from a user selected data source,

wherein each of the data streams includes a plurality of attributes which categorize a corresponding data stream'"'"'s total volume;

responsive to a selection of one or more attributes of a respective data stream, displaying, on a user interface, the data signals which correspond to the selected attributes using a time series plot;

determining correlations between the data streams based on calculating a correlation score for each correlation between (1) each selected attribute in the first data stream and (2) each selected attribute in the second data stream,

wherein the correlation score is calculated at least in part based on permutations of different lengths of time shifts in which the data signals take place; and

displaying, on the user interface, correlation scores adjacent to their respective attributes.

2. The method of claim 1, further comprising:

based on the correlation, employing a predictive algorithm to predict a future trend of the second series of the data signals in the second data stream; and

verifying the prediction by comparing the predicted future trend with real data signals as the real data signals arrive from the second data stream.

3. The method of claim 2, further comprising:

automatically adjusting the predictive algorithm based on results from the verification, wherein the automatic adjustment is governed by an unsupervised machine learning mechanism.

4. The method of claim 3, further comprising:

performing analysis to determine which one or more of the determined correlations are most predictive of changes to the second stream.

5. The method of claim 3, wherein the unsupervised machine learning mechanism comprises an implementation of a genetic algorithm.

6. The method of claim 1, further comprising:

monitoring subsequent data signals from the data streams to detect whether the determined correlations are affected; and

performing an action when the determined correlations are affected by the subsequent data signals.

7. The method of claim 6, wherein the action comprises generating a notification to a user.

8. The method of claim 1, further comprising:

importing a third data stream comprising a third time series of the data signals from another user selected data source.

9. The method of claim 1, wherein the one or more social networks comprise web-based, computer-facilitated platforms which primarily provide services for their users to build or maintain social relations with other users.

10. The method of claim 1, wherein the user selected data source comprises one or more of:

sales results for products, stock prices for companies, polling results, website access statistics, transaction data, stock trading volume, application usage statistics, macro-economic indicators, customer support statistics, inbound call center or sales statistics, or business financial metrics for a company.

11. The method of claim 1, wherein the conversations include one or more topics, and wherein the first data stream'"'"'s attributes include a volume of conversations corresponding to each topic.

12. The method of claim 11, wherein the first data stream'"'"'s attributes further include (1) a volume of conversations with positive sentiments for at least one topic, and (2) a volume of conversations with negative sentiments for the at least one topic.

13. The method of claim 11, wherein the first data stream'"'"'s attributes further include a volume of each gender which generated the conversations corresponding to the at least one topic.

14. The method of claim 11, wherein the first data stream'"'"'s attributes further include one or more of:

(1) an identity of network source;

(2) a volume of users;

(3) an ethnicity of users who generated corresponding data signals;

or (4) a volume of how many users interacted with corresponding data signals.

15. A server which reveals correlations in data signals, the server comprising:

a processor; and

a memory unit having instructions stored thereon which when executed by the processor, causes the processor to;

automatically import a first data stream comprising a first time series of the data signals from conversations occurred within one or more social networks;

import a second data stream comprising a second time series of the data signals from a user selected data source,

wherein each of the data streams includes a plurality of attributes which categorize a corresponding data stream'"'"'s total volume;

responsive to a selection of one or more attributes of a respective data stream, display, on a user interface, the data signals which correspond to the selected attributes using a time series plot;

determine correlations between the data streams based on calculating a correlation score for each correlation between (1) each selected attribute in the first data stream and (2) each selected attribute in the second data stream,

wherein the correlation score is calculated at least in part based on permutations of different lengths of time shifts in which the data signals take place; and

display, on the user interface, correlation scores adjacent to their respective attributes.

16. The server of claim 15, wherein the processor is further caused by the instructions to:

based on the correlation, employ a predictive algorithm to predict a future trend of the second series of the data signals in the second data stream; and

verify the prediction by comparing the predicted future trend with real data signals as the real data signals arrive from the second data stream.

17. The server of claim 16, wherein the processor is further caused by the instructions to:

automatically adjust the predictive algorithm based on results from the verification, wherein the automatic adjustment is governed by an unsupervised machine learning mechanism.

18. The server of claim 17, wherein the processor is further caused by the instructions to:

perform analysis to determine which one or more of the determined correlations are most predictive of changes to the second stream.

19. The server of claim 17, wherein the unsupervised machine learning mechanism comprises an implementation of a genetic algorithm.

20. The server of claim 15, wherein the processor is further caused by the instructions to:

monitor subsequent data signals from the data streams to detect whether the determined correlations are affected;

perform an action when the determined correlations are affected by the subsequent data signals.

21. The server of claim 20, wherein the action comprises generating a notification to a user.

22. The server of claim 15, wherein the processor is further caused by the instructions to:

import a third data stream comprising a third time series of the data signals from another user selected data source.

23. The server of claim 15, wherein the one or more social networks comprise web-based, computer-facilitated platforms which primarily provide services for their users to build or maintain social relations with other users.

24. The server of claim 15, wherein the user selected data source comprises one or more of:

sales results for products, stock prices for companies, polling results, website access statistics, transaction data, stock trading volume, application usage statistics, macro-economic indicators, customer support statistics, inbound call center or sales statistics, or business financial metrics for a company.

25. The server of claim 15, wherein the conversations include one or more topics, and wherein the first data stream'"'"'s attributes include a volume of conversations corresponding to each topic.

26. The server of claim 25, wherein the first data stream'"'"'s attributes further include (1) a volume of conversations with positive sentiments for at least one topic, and (2) a volume of conversations with negative sentiments for the at least one topic.

27. The server of claim 25, wherein the first data stream'"'"'s attributes further include a volume of each gender which generated the conversations corresponding to the at least one topic.

28. A system which reveals correlations in data signals, the system comprising:

means for automatically importing a first data stream comprising a first time series of the data signals from conversations occurred within one or more social networks;

means for importing a second data stream comprising a second time series of the data signals from a user selected data source,

wherein each of the data streams includes a plurality of attributes which categorize a corresponding data stream'"'"'s total volume;

means for, responsive to a selection of one or more attributes of a respective data stream, displaying, on a user interface, the data signals which correspond to the selected attributes using a time series plot;

means for determining correlations between the data streams based on calculating a correlation score for each correlation between (1) each selected attribute in the first data stream and (2) each selected attribute in the second data stream,

wherein the correlation score is calculated at least in part based on permutations of different lengths of time shifts in which the data signals take place; and

means for displaying, on the user interface, correlation scores adjacent to their respective attributes.