Interaction-based visualization to augment user experience

1. A method for providing real-time recommendations to a user, the method comprising:

capturing user interactions with a data visualization, wherein the user interactions comprise images captured as the user interacts with the data visualization;

building image stacks of the user interactions, wherein the image stacks of the user interactions are built from sequences of the user interactions captured over time;

generating embeddings based on the image stacks of the user interactions;

finding clusters of the embeddings having similar properties; and

making the real-time recommendations to the user based on the clusters of the embeddings having the similar properties.

2. The method of claim 1 , further comprising:

storing the embeddings and associated metadata in a repository.

3. The method of claim 1 , wherein an embedding is generated for each user interaction in the sequences of the user interactions.

4. The method of claim 1 , wherein the data visualization is selected from the group consisting of: a chart, a graph, a table, a report, video, and combinations thereof.

5. The method of claim 1 , wherein the user interactions comprise overlay images, and wherein the image stacks of the user interactions are built from at least one sequence of the overlay images captured over time.

6. The method of claim 5 , wherein an embedding is generated for each overlay image in the at least one sequence of the overlay images.

7. The method of claim 1 , wherein the embeddings are generated from a neural network.

8. The method of claim 7 , further comprising:

pre-training the neural network using positive embeddings and negative embeddings.

9. The method of claim 1 , further comprising:

computing similarity scores between all of the embeddings; and

forming the clusters of the embeddings having the similar properties using the similarity scores.

10. The method of claim 9 , further comprising:

determining which embedding has a maximum number of nearest neighbors.

11. The method of claim 10 , further comprising:

selecting the embedding having the maximum number of nearest neighbors;

retrieving metadata associated with the embedding having the maximum number of nearest neighbors; and

extracting properties from the metadata associated with the embedding having the maximum number of nearest neighbors.

12. The method of claim 1 , wherein the real-time recommendations are selected from the group consisting of: at least one interaction from the image stacks of the user interactions, metadata associated with the at least one interaction, and combinations thereof.

13. The method of claim 12 , wherein the real-time recommendations comprise the at least one interaction from the image stacks of the user interactions, and wherein the method further comprises:

displaying to the user a sequence of the user interactions that led to the at least one interaction.

14. The method of claim 12 , wherein the real-time recommendations comprise the metadata associated with the at least one interaction, and wherein the method further comprises:

displaying to the user a summary of the metadata.

15. The method of claim 1 , wherein the real-time recommendations comprise different types of data visualizations than the user is currently viewing.

16. A non-transitory computer program product for providing real-time recommendations to a user, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to:

capture user interactions with a data visualization, wherein the user interactions comprise images captured as the user interacts with the data visualization;

build image stacks of the user interactions, wherein the image stacks of the user interactions are built from sequences of the user interactions captured over time;

generate embeddings based on the image stacks of the user interactions;

find clusters of the embeddings having similar properties; and

make the real-time recommendations to the user based on the clusters of the embeddings having the similar properties.

17. The non-transitory computer program product of claim 16 , wherein the program instructions further cause the computer to:

store the embeddings and associated metadata in a repository.

18. The non-transitory computer program product of claim 16 , wherein the program instructions further cause the computer to:

compute similarity scores between all of the embeddings; and

form the clusters of the embeddings having the similar properties using the similarity scores.

19. The non-transitory computer program product of claim 18 , wherein the program instructions further cause the computer to:

determine which embedding has a maximum number of nearest neighbors.

20. The non-transitory computer program product of claim 19 , wherein the program instructions further cause the computer to:

select the embedding having the maximum number of nearest neighbors;

retrieve metadata associated with the embedding having the maximum number of nearest neighbors; and

extract properties from the metadata associated with the embedding having the maximum number of nearest neighbors.