Machine learning-powered framework to transform overloaded text documents

1. A method to transform an overloaded text document into a plurality of pages, the method comprising:

accessing the overloaded text document;

accessing user preferences from a user profile that indicate situations when overloaded text documents are to be altered;

determining that an amount of text on the overloaded text document transgresses a threshold that triggers a process to transform the overloaded text document into the plurality of pages;

in response to the determining and based on the user preferences, analyzing, by one or more hardware processors, the overloaded text document to identify segmentation data, the segmentation data indicating a plurality of segments derived from the overloaded text document, each segment of the plurality of segments to be incorporated into a different page of the plurality of pages of a template that is selected based on the user preferences, number of pages needed, and an amount of text in each segment to be included on each page;

generating the plurality of pages by incorporating each segment of the plurality of segments into a different corresponding page of the plurality of pages and providing a transition between each of the plurality of pages; and

causing display of the plurality of pages as a recommendation.

2. The method of claim 1 , wherein the template is selected based on one or more of a time of day, a day of week, a location, or a client endpoint based on the user preferences.

3. The method of claim 1 , wherein the transition comprises a morph transition between each page of the plurality of pages.

4. The method of claim 1 , wherein the overloaded text document comprises an overloaded slide and the generating the plurality of pages comprises generating a plurality of slides from the overloaded slide.

5. The method of claim 1 , further comprising:

receiving an indication of acceptance or rejection of the recommendation; and

using the indication as feedback for adjusting user preferences or retraining a machine-learning model.

6. The method of claim 1 , wherein the analyzing the overloaded text document to identify segmentation data comprises using a segmentation model to segment the overloaded text document into smaller parts and identify a number of segments that the overloaded text document will be split into.

7. The method of claim 1 , wherein the analyzing the overloaded text document to identify segmentation data comprises analyzing the overloaded text document for structure signals, the structure signals including one or more of numbering, end of a paragraph, bullet points, a new line, or a period symbol.

8. The method of claim 1 , further comprising using a summarization model to extract key parts from the overloaded text document for use as a title for each segment.

9. The method of claim 1 , further comprising:

accessing a plurality of candidate templates, each candidate template comprising a plurality of template pages having a different background element that shares a common theme; and

ranking the plurality of candidate templates based, in part, on the segmentation data to determine top ranked candidate templates.

10. The method of claim 9 , wherein:

the generating the plurality of pages comprises generating top candidate page sets using a predetermined number of the top ranked candidate templates by incorporating each segment of the plurality of segments into a different corresponding page of the plurality of template pages for each of the top ranked candidate templates; and

the causing display of the plurality of pages comprises causing display of the top candidate page sets as the recommendation.

11. The method of claim 1 , further comprising generating a candidate template, the generating the candidate template comprises:

selecting a large image; and

based on design constraints, positioning cropping boxes on different locations of the large image, each location of each cropping box on the image resulting in a different background of a page of the plurality of pages.

12. The method of claim 1 , further comprising generating a candidate template, the generating the candidate template comprises:

selecting a blueprint family; and

based on design constraints, selecting a subset of the blueprint family and applying an order to the subset of the blueprint family, the subset of the blueprint family corresponding to different backgrounds of the plurality of pages.

13. A system to transform an overloaded text document into a plurality of pages, the system comprising:

one or more hardware processors; and

a memory storing instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising:

accessing the overloaded text document;

accessing user preferences from a user profile that indicate situations when overloaded text documents are to be altered;

determining that an amount of text on the overloaded text document transgresses a threshold that triggers a process to transform the overloaded text document into the plurality of pages;

in response to the determining and based on the user preferences, analyzing, by one or more hardware processors, the overloaded text document to identify segmentation data, the segmentation data indicating a plurality of segments derived from the overloaded text document, each segment of the plurality of segments to be incorporated into a different page of the plurality of pages of a template that is selected based on the user preferences, number of pages needed, and an amount of text in each segment to be included on each page;

generating the plurality of pages by incorporating each segment of the plurality of segments into a different corresponding page of the plurality of pages and providing a transition between each of the plurality of pages; and

causing display of the plurality of pages as a recommendation.

14. The system of claim 13 , wherein the template is selected based on one or more of a time of day, a day of week, a location, or a client endpoint based on the user preferences.

15. The system of claim 13 , wherein the transition comprises a morph transition between each page of the plurality of pages.

16. The system of claim 13 , wherein the overloaded text document comprises an overloaded slide and the generating the plurality of pages comprises generating a plurality of slides from the overloaded slide.

17. The system of claim 13 , wherein the operations further comprise:

receiving an indication of acceptance or rejection of the recommendation; and

using the indication as feedback for adjusting user preferences or retraining a machine-learning model.

18. The system of claim 13 , wherein the analyzing the overloaded text document to identify segmentation data comprises using a segmentation model to segment the overloaded text document into smaller parts and identify a number of segments that the overloaded text document will be split into.

19. The system of claim 13 , wherein the analyzing the overloaded text document to identify segmentation data comprises determining context associated with the overloaded text document to identify a change in context, the change in context indicating a different segment.

20. A computer-storage medium comprising instructions which, when executed by one or more hardware processors of a machine, cause the machine to perform operations comprising:

accessing the overloaded text document;

accessing user preferences from a user profile that indicate situations when overloaded text documents are to be altered;

determining that an amount of text on the overloaded text document transgresses a threshold that triggers a process to transform the overloaded text document into the plurality of pages;

in response to the determining and based on the user preferences, analyzing, by one or more hardware processors, the overloaded text document to identify segmentation data, the segmentation data indicating a plurality of segments derived from the overloaded text document, each segment of the plurality of segments to be incorporated into a different page of the plurality of pages of a template that is selected based on the user preferences, number of pages needed, and an amount of text in each segment to be included on each page;

generating the plurality of pages by incorporating each segment of the plurality of segments into a different corresponding page of the plurality of pages and providing a transition between each of the plurality of pages; and

causing display of the plurality of pages as a recommendation.