Method and system for automatic formality transformation

1. A method, implemented on a machine having at least one processor, storage, and a communication platform connected to a network for transforming a formality level of a text item in an online communication, the method comprising:

receiving a request for transforming a formality level of a text item in an online communication;

obtaining a current formality level of the text item, wherein the current formality level represents a current degree of formality of the text item;

obtaining contextual information with respect to the online communication, wherein the contextual information is indicative of a role of the text item in the online communication;

selecting, from a plurality of target granularity levels, a target granularity level at which the text item is to be transformed based on an estimated purpose of the online communication;

determining a target formality level for the text item based on the estimated purpose, wherein the target formality level represents a targeted degree of formality for the text item;

transforming, at the granularity level, the text item having the current formality level to a transformed text item having the target formality level based on a formality transformation model trained using training contextual information indicating purposes associated with online communications, wherein the transformed text item has a same literal meaning as the text item; and

providing the transformed text item as a response to the request.

2. The method of claim 1 , further comprising:

generating a naive transformation model based on a larger corpus of artificially generated training data;

adjusting the naive transformation model to generate the formality transformation model further trained based on a smaller corpus of manually built training data, wherein the text item is automatically transformed to the transformed text item based on the formality transformation model.

3. The method of claim 2 , wherein:

both the artificially generated training data and the manually built training data include parallel data associated with a plurality of training text items; and

with respect to each of the plurality of training text items, the parallel data include different versions of the training text item such that the different versions of the training text item have a same literal meaning but different formality levels.

4. The method of claim 1 , further comprising:

obtaining one or more linguistic features extracted from the text item, wherein the text item is further transformed to the transformed text item based on the one or more linguistic features and the contextual information.

5. The method of claim 1 , further comprising:

updating a user profile of a user associated with the online communication based on the transformed text item such that the user profile indicates that the target formality level is one formality level preferred by the user.

6. The method of claim 1 , wherein:

the current formality level includes one or more first real values representing first degrees of formality of the text item in accordance with the granularity level; and

the target formality level includes one or more second real values representing second degrees of formality of the transformed text item in accordance with the granularity level.

7. The method of claim 1 , further comprising:

selecting the formality transformation model based on the target granularity level, wherein the formality transformation model comprises a machine learning model trained, and wherein the machine learning model is further trained with parallel textual data.

8. A method, implemented on a machine having at least one processor, storage, and a communication platform connected to a network for transforming a formality level of a text item to be recommended to an online user, the method comprising:

receiving a request for transforming a formality level of a text item to be recommended to an online user;

obtaining a current formality level of the text item, wherein the current formality level represents a current degree of formality of the text item;

obtaining contextual information with respect to the online communication, wherein the contextual information is indicative of a role of the text item in the online communication;

selecting, from a plurality of target granularity levels, a target granularity level at which the text item is to be transformed based on a formality preference of the online user and an estimated purpose of the online communication;

determining a target formality level for the text item based on the estimated purpose of the text item and personal information of the online user, wherein the target formality level represents a targeted degree of formality for the text item, and the personal information comprises the formality preference of the online user;

transforming, at the granularity level, the text item having the current formality level to a transformed text item having the target formality level based on a formality transformation model trained using training contextual information indicating purposes associated with online communications, wherein the transformed text item has a same literal meaning as the text item; and

providing the transformed text item as a response to the request.

9. The method of claim 8 , further comprising:

determining a user identity (ID) of the online user; and

retrieving a user profile associated with the user ID, wherein the user profile comprises the formality preference of the online user.

10. The method of claim 8 , wherein the formality preference of the online user is determined based on historical online behaviors of the online user.

11. A system having at least one processor, storage, and a communication platform connected to a network for transforming a formality level of a text item in an online communication, the system comprising:

a formality transformation request analyzer configured for receiving a request for transforming a formality level of a text item in an online communication;

a formality level information obtainer configured for obtaining a current formality level of the text item, wherein the current formality level represents a current degree of formality of the text item and obtaining contextual information with respect to the online communication, wherein the contextual information is indicative of a role of the text item in the online communication;

a transformation granularity determiner configured for selecting, from a plurality of target granularity levels, a target granularity level at which the text item is to be transformed based on an estimated purpose of the online communication;

a target formality determiner configured for determining a target formality level for the text item based on the estimated purpose, wherein the target formality level represents a targeted degree of formality for the text item; and

a formality transformer configured for:

transforming, at the granularity level, the text item having the current formality level to a transformed text item having the target formality level based on a formality transformation model trained using training contextual information indicating purposes associated with online communications, and

providing the transformed text item as a response to the request, wherein the transformed text item has a same literal meaning as the text item.

12. The system of claim 11 , further comprising:

a naive transformation model generator configured for generating a naive transformation model based on a larger corpus of artificially generated training data; and

a formality transformation model generator configured for adjusting the naive transformation model to generate the formality transformation model based on a smaller corpus of manually built training data, wherein the text item is automatically transformed to the transformed text item based on the formality transformation model.

13. The system of claim 12 , wherein:

both the artificially generated training data and the manually built training data include parallel data associated with a plurality of training text items; and

with respect to each of the plurality of training text items, the parallel data include different versions of the training text item such that the different versions of the training text item have a same literal meaning but different formality levels.

14. The system of claim 11 , wherein the formality level information obtainer is further configured for:

obtaining one or more linguistic features extracted from the text item, wherein the text item is transformed to the transformed text item based on the one or more linguistic features and the contextual information.

15. The system of claim 11 , further comprising:

a user profile generator/updater configured for updating a user profile of a user associated with the online communication based on the transformed text item such that the user profile indicates that the target formality level is one formality level preferred by the user.

16. A non-transitory machine-readable medium having information recorded thereon for transforming a formality level of a text item in an online communication, wherein the information, when read by the machine, causes the machine to perform the following:

receiving a request for transforming a formality level of a text item in an online communication;

obtaining a current formality level of the text item, wherein the current formality level represents a current degree of formality of the text item;

obtaining contextual information with respect to the online communication, wherein the contextual information is indicative of a role of the text item in the online communication;

selecting, from a plurality of target granularity levels, a target granularity level at which the text item is to be transformed based on an estimated purpose of the online communication;

determining a target formality level for the text item based on the estimated purpose, wherein the target formality level represents a targeted degree of formality for the text item;

transforming, at the granularity level, the text item having the current formality level to a transformed text item having the target formality level based on a formality transformation model trained using training contextual information indicating purposes associated with online communications, wherein the transformed text item has a same literal meaning as the text item; and

providing the transformed text item as a response to the request.

17. The medium of claim 16 , wherein the information, when read by the machine, further causes the machine to perform the following:

generating a naive transformation model based on a larger corpus of artificially generated training data; and

adjusting the naive transformation model to generate the formality transformation model based on a smaller corpus of manually built training data, wherein the text item is automatically transformed to the transformed text item based on the formality transformation model.

18. The medium of claim 17 , wherein:

both the artificially generated training data and the manually built training data include parallel data associated with a plurality of training text items; and

with respect to each of the plurality of training text items, the parallel data include different versions of the training text item such that the different versions of the training text item have a same literal meaning but different formality levels.

19. The medium of claim 16 , wherein the information, when read by the machine, further causes the machine to perform the following:

obtaining one or more linguistic features extracted from the text item, wherein the text item is transformed to the transformed text item based on the one or more linguistic features and the contextual information.

20. The medium of claim 16 , wherein the information, when read by the machine, further causes the machine to perform the following:

updating a user profile of a user associated with the online communication based on the transformed text item such that the user profile indicates that the target formality level is one formality level preferred by the user.