Prediction and isolation of patterns across datasets

• US 9,069,880 B2
• Filed: 03/16/2012
• Issued: 06/30/2015
• Est. Priority Date: 03/16/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Trim, Samuel T., Snow, Brad
• Primary Examiner(s): Le, Miranda
• Application Number:
  US13/421,905
• Publication Number:
  US 20130246463A1
• Time in Patent Office: 1,201 Days
• Field of Search: 707/776, 707/737, 707/740, 707/780
• US Class Current: 1/1
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/9024 Graphs; Linked lists G06F16/9027 takes precedence
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: H04L 41/0631 using root cause analysis; using analysis of correlation between notifications, alarms or events based on decision criteria, e.g. hierarchy, tree or time analysis
• CPC Class Codes: H04L 41/0677 Localisation of faults
• CPC Class Codes: H04L 41/147 for predicting network behaviour
• CPC Class Codes: H04L 67/00 Network arrangements or protocols for supporting network services or applications user-to-user messaging H04L51/00; network arrangements, protocols or services for supporting real-time applications in data packet communications networks H04L65/00
• CPC Class Codes: H04L 67/52 specially adapted for the location of the user terminal

Abstract

Various embodiments pertain to techniques for predicting and isolating patterns or trends across datasets. In various embodiments, one or more Q-entities are extracted from a data seed, associated with one or more dimensions, and classified into one or more clusters for each dimension with which it is associated. In some embodiments, a Q-entity can exist in more than one dimension and/or more than one cluster within a dimension. Once information from the data seed is associated with a dimension and cluster, frequency analysis can be utilized to ascertain a pattern or trend in the data. In various embodiments, additional data can be processed, added to the dimensions and clusters, and frequency analysis can be performed on the updated dataset to provide additional information on the pattern or trend.

Claims

1. A method comprising:

receiving a data seed for use in identifying a trend across data in a network-based environment;

extracting a Q-entity from the data seed;

obtaining additional Q-entities based on the Q-entity extracted from the data seed, wherein a subsequent data seed is utilized to obtain the additional Q-entities, the subsequent data seed comprising a web page to which each of a plurality of hyperlinks redirect a web browser;

analyzing each Q-entity to determine when the Q-entity belongs to more than one dimension;

analyzing, using a relational graph model having at least one Q-dimension having at least one Q-cluster, the Q-entity effective to associate one of the Q-entities with the at least one Q-cluster and the at least one Q-dimension of the relational graph; and

identifying, based on a frequency analysis of the Q-cluster in the relational graph model to determine clustering patterns that signify relational ties between Q-entities that are used to identify the trend, wherein the trend is identified according to an expression;

2. The method of claim 1, the at least one Q-dimension comprising one or more of a visual dimension, a sound dimension, or an emotion dimension.

3. The method of claim 2, the at least one Q-cluster of the visual dimension comprising one or more of colors, location/environment, or visual media.

4. The method of claim 2, the at least one Q-cluster of the sound dimension comprising one or more of oral history, music, or language.

5. The method of claim 2, the at least one Q-cluster of the emotion dimension comprising one or more of hopes and desires, general mood, or fear.

6. The method of claim 1, extracting at least one Q-entity comprising extracting one or more:

keywords, words used to express emotion, a username on a social network site, a date and/or time a message was posted, or a user'"'"'s location according to a user'"'"'s social network profile.

7. The method of claim 1, analyzing the Q-entity comprising classifying the Q-entity into the at least one Q-cluster using fuzzy clustering.

8. The method of claim 7, using fuzzy clustering comprising determining a strength of a Q-entity in a cluster according to an expression:

9. One or more computer-readable storage media comprising instructions that are executable to cause a device to perform a process comprising:

extracting, from a data seed, one or more Q-entities;

obtaining additional Q-entities based on the one or more Q-entities extracted from the data seed, wherein a subsequent data seed is utilized to obtain the additional Q-entities, the subsequent data seed comprising a web page to which each of a plurality of hyperlinks redirect a web browser;

analyzing the Q-entity to determine when the Q-entity belongs to more than one dimension;

analyzing using a relational graph model having a plurality of Q-dimensions each having a plurality of Q-clusters, each of the Q-entities effective to associate each Q-entity with a Q-cluster and Q-dimension; and

identifying, based on frequency analysis of the Q-clusters in the relational graph model to determine clustering patterns that signify relational ties between Q-entities that are used to identify a trend in a web environment to which the data seed pertains, wherein the trend is identified according to an expression;

10. The one or more computer-readable storage media of claim 9, extracting one or more Q-entities comprising extracting one or more:

keywords, words used to express emotion, a username on a social network site, a date and/or time a message was posted, or a user'"'"'s location according to a user'"'"'s social network profile.

11. The one or more computer-readable storage media of claim 9, the data seed comprising a post from a microblogging service, a post from a social network service, a result from a search engine, information obtained from a trend map service, information obtained from website analytics, or information from a publicly available database.

12. The one or more computer-readable storage media of claim 9, the plurality of Q-dimensions comprising a visual dimension, a sound dimension, and an emotion dimension, wherein each of the plurality of Q-dimensions is assigned a value from one to three.

13. The one or more computer-readable storage media of claim 12, the plurality of Q-clusters of the visual dimension comprising colors, location/environment, and visual media.

14. The one or more computer-readable storage media of claim 12, the plurality of Q-clusters of the sound dimension comprising oral history, music, and language.

15. The one or more computer-readable storage media of claim 12, the plurality of Q-clusters of the visual dimension comprising hopes and desires, general mood, and fears.

16. A device comprising:

one or more processors;

one or more computer-readable storage media;

one or more modules embodied on the one or more computer-readable storage media and executable under the influence of the one or more processors, the one or more modules configured to;

extract, from a data seed from a web environment, Q-entities;

obtain additional Q-entities based on the Q-entity extracted from the data seed, wherein a subsequent data seed is utilized to obtain the additional Q-entities, the subsequent data seed comprising a web page to which each of a plurality of hyperlinks redirect a web browser;

analyze each Q-entity to determine when the Q-entity belongs to more than one dimension;

analyze each Q-entity using a relational graph model having three Q-dimensions, each Q-dimension having three Q-clusters; and

identify a trend present in the web environment according to related concepts in one or more of the Q-clusters by determining clustering patterns that signify relational ties between the Q-entities, wherein the trend is identified according to an expression;

17. The device of claim 16, wherein the one or more modules being configured to analyze each Q-entity comprises the one or more modules being configured to:

assign a unique value from one to three to each of the Q-dimensions and to each of the Q-clusters within a dimension; and

utilize a fuzzy clustering algorithm to determine the one or more Q-dimensions and the one or more Q-clusters to which the Q-entity belongs, wherein utilizing the fuzzy clustering algorithm comprises using the unique value to represent the Q-dimension and Q-cluster to which it is assigned.