Insight determination and explanation in multi-dimensional data sets

1. A computer program product for reason determination for insights gleaned from multi-dimensional data, based on directional properties of the insights, the computer program product comprising:

a non-transitory computer-readable medium having program code embodied therewith, the program code executable by one or more computer processors to:

determine one or more insights comprising a subset of cells in a contingency table, each of the one or more insights satisfying a relevance criterion and having an associated directional property specifying whether the respective insight occurs more frequently or less frequently than a statistically expected frequency for the respective insight, wherein satisfying the relevance criterion comprises satisfying a threshold deviation from the statistically expected frequency for the respective insight; and

determine, based on the directional property of a first insight of the one or more insights, a reason pertaining to the first insight, the reason comprising a member that is identified as characterizing a rationale underlying the first insight, wherein determining the reason pertaining to the first insight comprises, upon determining that a baseline ratio and a candidate ratio satisfy a reason criterion, designating the identified member from the selected dimension, as the reason pertaining to the first insight, wherein an indication is output that the identified member is designated as the reason pertaining to the first insight;

wherein the reason criterion is satisfied upon at least one of a first condition and a second condition; the first condition comprising: (i) the first insight having a positive directional property; and (ii) the candidate ratio being greater than the baseline ratio by at least a threshold magnitude; the second condition comprising: (i) the first insight having a negative directional property; and (ii) the candidate ratio being less than the baseline ratio by at least the threshold magnitude.

2. The computer program product of claim 1 , wherein the program is further executable to:

access the contingency table generated based on a set of data snippets collectively reflecting a plurality of dimensions of multi-dimensional data, wherein the contingency table comprises a plurality of cells, each of the plurality of dimensions having a plurality of members, each of the plurality of cells representing a distinct combination of members selected from a subset of different dimensions of the plurality of dimensions, each of the plurality of cells having a frequency property characterizing a frequency of occurrence, in the set of data snippets, of the distinct combination represented by the respective cell, each of the plurality of cells further having an associated relevance property;

wherein the identified member is from a dimension selected from a set difference between the plurality of dimensions and the subset of different dimensions represented by the first insight.

3. The computer program product of claim 2 , wherein the distinct combination of members represented by the first insight is reflected by a subset of the set of data snippets, wherein determining the reason pertaining to the first insight further comprises:

determining the baseline ratio, which comprises a ratio of: (i) any data snippets, selected from the set of data snippets, that reflect the identified member from the selected dimension; to (ii) the set of data snippets; and

determining the candidate ratio, which comprises a ratio of: (i) any data snippets, selected from the subset of data snippets, that further reflect the identified member; to (ii) the subset of data snippets.

4. The computer program product of claim 3 , wherein having a positive directional property comprises the directional property of the first insight specifying that the first insight occurs more frequently than the statistically expected frequency for the first insight;

wherein having a negative directional property comprises the directional property of the first insight specifying that the first insight occurs less frequently than the statistically expected frequency for the first insight.

5. A computer system for reason determination for insights gleaned from multi-dimensional data, based on directional properties of the insights, the computer system comprising:

one or more computer processors, one or more computer-readable memories and one or more computer-readable, tangible storage devices; and

program instructions, stored on at least one of the one or more computer-readable, tangible storage devices for execution by at least one of the one or more computer processors via at least one of the one or more computer-readable memories, to:

determine one or more insights comprising a subset of cells in a contingency table, each of the one or more insights satisfying a relevance criterion and having an associated directional property specifying whether the respective insight occurs more frequently or less frequently than a statistically expected frequency for the respective insight, wherein satisfying the relevance criterion comprises satisfying a threshold deviation from the statistically expected frequency for the respective insight; and

determine, based on the directional property of a first insight of the one or more insights, a reason pertaining to the first insight, the reason comprising a member that is identified as characterizing a rationale underlying the first insight, wherein determining the reason pertaining to the first insight comprises, upon determining that a baseline ratio and a candidate ratio satisfy a reason criterion, designating the identified member from the selected dimension, as the reason pertaining to the first insight, wherein an indication is output that the identified member is designated as the reason pertaining to the first insight;

wherein the reason criterion is satisfied upon at least one of a first condition and a second condition; the first condition comprising: (i) the first insight having a positive directional property; and (ii) the candidate ratio being greater than the baseline ratio by at least a threshold magnitude; the second condition comprising: (i) the first insight having a negative directional property; and (ii) the candidate ratio being less than the baseline ratio by at least the threshold magnitude.

6. The computer system of claim 5 , further comprising program instructions to:

access the contingency table generated based on a set of data snippets collectively reflecting a plurality of dimensions of multi-dimensional data, wherein the contingency table comprises a plurality of cells, each of the plurality of dimensions having a plurality of members, each of the plurality of cells representing a distinct combination of members selected from a subset of different dimensions of the plurality of dimensions, each of the plurality of cells having a frequency property characterizing a frequency of occurrence, in the set of data snippets, of the distinct combination represented by the respective cell, each of the plurality of cells further having an associated relevance property;

wherein the identified member is from a dimension selected from a set difference between the plurality of dimensions and the subset of different dimensions represented by the first insight.

7. The computer system of claim 6 , wherein the distinct combination of members represented by the first insight is reflected by a subset of the set of data snippets, wherein determining the reason pertaining to the first insight comprises:

determining the baseline ratio, which comprises a ratio of: (i) any data snippets, selected from the set of data snippets, that reflect the identified member from the selected dimension; to (ii) the set of data snippets; and

determining the candidate ratio, which comprises a ratio of: (i) any data snippets, selected from the subset of data snippets, that further reflect the identified member; to (ii) the subset of data snippets.

8. The computer system of claim 7 , wherein having a positive directional property comprises the directional property of the first insight specifying that the first insight occurs more frequently than the statistically expected frequency for the first insight;

wherein having a negative directional property comprises the directional property of the first insight specifying that the first insight occurs less frequently than the statistically expected frequency for the first insight.

9. The computer system of claim 8 , wherein each member in each distinct combination of members represented by the plurality of cells is from a different dimension, wherein the computer system further comprises program instructions to:

generate the contingency table based on the set of data snippets, including determining the frequency property specified in each cell of the contingency table; and

determine, based on one or more statistical tests, a relevance property of each cell in the contingency table.

10. The computer program product of claim 1 , wherein having a positive directional property comprises the directional property of the first insight specifying that the first insight occurs more frequently than the statistically expected frequency for the first insight.

11. The computer program product of claim 1 , wherein having a negative directional property comprises the directional property of the first insight specifying that the first insight occurs less frequently than the statistically expected frequency for the first insight.

12. The computer program product of claim 1 , wherein determining the reason pertaining to the first insight further comprises:

determining the baseline ratio, which comprises a ratio of: (i) any data snippets, selected from the set of data snippets, that reflect the identified member from the selected dimension; to (ii) the set of data snippets.

13. The computer program product of claim 1 , wherein determining the reason pertaining to the first insight further comprises:

determining the candidate ratio, which comprises a ratio of: (i) any data snippets, selected from the subset of data snippets, that further reflect the identified member; to (ii) the subset of data snippets.

14. The computer program product of claim 1 , wherein at least one relevance property in the contingency table is generated without requiring user intervention.

15. The computer program product of claim 1 , wherein a plurality of distinct reasons pertaining to the first insight, including the reason, is designated.

16. The computer system of claim 5 , wherein having a positive directional property comprises the directional property of the first insight specifying that the first insight occurs more frequently than the statistically expected frequency for the first insight.

17. The computer system of claim 5 , wherein having a negative directional property comprises the directional property of the first insight specifying that the first insight occurs less frequently than the statistically expected frequency for the first insight.

18. The computer system of claim 5 , wherein determining the reason pertaining to the first insight further comprises:

determining the baseline ratio, which comprises a ratio of: (i) any data snippets, selected from the set of data snippets, that reflect the identified member from the selected dimension; to (ii) the set of data snippets.

19. The computer system of claim 5 , wherein determining the reason pertaining to the first insight further comprises:

determining the candidate ratio, which comprises a ratio of: (i) any data snippets, selected from the subset of data snippets, that further reflect the identified member; to (ii) the subset of data snippets.

20. The computer system of claim 5 , wherein at least one relevance property in the contingency table is generated without requiring user intervention.

21. The computer system of claim 5 , wherein a plurality of distinct reasons pertaining to the first insight, including the reason, is designated.

22. The computer system of claim 5 , wherein the multi-dimensional data comprises user-generated social media content.