Transformations of a user-interface modality of an application

1. A method, comprising:

identifying, with a data processing system, a first workflow segment corresponding to a user-interface (UI) modality of an application, wherein the application is developed to run on a predetermined data processing platform;

selecting, with the data processing system, at least one other workflow segment to transform the UI modality of the application, wherein the at least one other workflow segment is configured to perform, when executed on a different predetermined data processing platform, a function comparable to a function performable by the first workflow segment; and

transforming, using the selected at least one other workflow segment, the UI modality of the application to perform, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment;

wherein the at least one other workflow segment is selected from a plurality of alternative workflow segments that are semantically similar to the first workflow segment, and wherein the selecting is based on classifying the first workflow segment with a classification model trained using machine learning to map workflow segments and corresponding UI modalities of the application to different predetermined data processing platforms.

2. The method of claim 1 , further comprising generating a rank-ordered list of recommended workflow segments in response to selecting the at least one other workflow segment comprising a plurality of other workflow segments.

3. The method of claim 1 , further comprising transforming the UI modality of the application by automatically substituting a second code segment for a first code segment, wherein the first code segment corresponds to the first workflow segment, and wherein the second code segment corresponds to the at least one other workflow segment and performs, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment.

4. The method of claim 1 , further comprising predicting a domain of the application during an in-development phase using a prediction model trained with machine learning to predict the domain of the application based on a predetermined set of application features.

5. The method of claim 1 , further comprising generating a training corpus for training the classification model by extracting workflow semantics and corresponding UI modalities of the application from each of a plurality of platform-specific versions of previously developed applications.

6. The method of claim 1 , further comprising generating the plurality of alternative workflow segments by clustering semantically similar workflow segments with a classification model trained using machine learning to cluster the workflow segments based on a predetermined distance metric.

7. The method of claim 1 , further comprising determining the corresponding UI modalities of the application based on accessibility text of the plurality of alternative workflow segments.

8. A system, comprising:

computer hardware having at least one processor programmed to initiate executable operations, the executable operations including:

identifying a first workflow segment corresponding to a user-interface (UI) modality of an application, wherein the application is developed to run on a predetermined data processing platform;

selecting at least one other workflow segment to transform the UI modality of the application, wherein the at least one other workflow segment is configured to perform, when executed on a different predetermined data processing platform, a function comparable to a function performable by the first workflow segment; and

transforming, using the selected at least one other workflow segment, the UI modality of the application to perform, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment;

wherein the at least one other workflow segment is selected from a plurality of alternative workflow segments that are semantically similar to the first workflow segment, and wherein the selecting is based on classifying the first workflow segment with a classification model trained using machine learning to map workflow segments and corresponding UI modalities of the application to different predetermined data processing platforms.

9. The system of claim 8 , wherein the executable operations further include generating a rank-ordered list of recommended workflow segments in response to selecting the at least one other workflow segment comprising a plurality of other workflow segments.

10. The system of claim 8 , wherein the executable operations further include transforming the UI modality of the application by automatically substituting a second code segment for a first code segment, wherein the first code segment corresponds to the first workflow segment, and wherein the second code segment corresponds to the at least one other workflow segment and performs, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment.

11. The system of claim 8 , wherein the executable operations further include predicting a domain of the application during an in-development phase using a prediction model trained with machine learning to predict the domain of the application based on a predetermined set of application features.

12. The system of claim 8 , wherein the executable operations further include generating a training corpus for training the classification model by extracting workflow semantics and corresponding UI modalities of the application from each of a plurality of platform-specific versions of previously developed applications.

13. The system of claim 8 , wherein the executable operations further include generating the plurality of alternative workflow segments by clustering semantically similar workflow segments with a classification model trained using machine learning to cluster the workflow segments based on a predetermined distance metric.

14. A computer program product, comprising:

a computer readable storage medium having program code stored thereon, the program code executable by a computer to initiate operations including:

identifying a first workflow segment corresponding to a user-interface (UI) modality of an application, wherein the application is developed to run on a predetermined data processing platform;

selecting at least one other workflow segment to transform the UI modality of the application, wherein the at least one other workflow segment is configured to perform, when executed on a different predetermined data processing platform, a function comparable to a function performable by the first workflow segment; and

transforming, using the selected at least one other workflow segment, the UI modality of the application to perform, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment;

wherein the at least one other workflow segment is selected from a plurality of alternative workflow segments that are semantically similar to the first workflow segment, and wherein the selecting is based on classifying the first workflow segment with a classification model trained using machine learning to map workflow segments and corresponding UI modalities of the application to different predetermined data processing platforms.

15. The computer program product of claim 14 , wherein the operations further include generating a rank-ordered list of recommended workflow segments in response to selecting the at least one other workflow segment comprising a plurality of other workflow segments.

16. The computer program product of claim 14 , wherein the operations further include transforming the UI modality of the application by automatically substituting a second code segment for a first code segment, wherein the first code segment corresponds to the first workflow segment, and wherein the second code segment corresponds to the at least one other workflow segment and performs, when executed on the different predetermined data processing platform, the function comparable to the function performable by the first workflow segment.

17. The computer program product of claim 14 , wherein the operations further include predicting a domain of the application during an in-development phase using a prediction model trained with machine learning to predict the domain of the application based on a predetermined set of application features.

18. The computer program product of claim 14 , wherein the operations further include generating a training corpus for training the classification model by extracting workflow semantics and corresponding UI modalities of the application from each of a plurality of platform-specific versions of previously developed applications.

19. The computer program product of claim 14 , wherein the operations further include generating the plurality of alternative workflow segments by clustering semantically similar workflow segments with a classification model trained using machine learning to cluster the workflow segments based on a predetermined distance metric.

20. The computer program product of claim 14 , wherein the operations further include determining the corresponding UI modalities of the application based on accessibility text of the plurality of alternative workflow segments.