Word attribution prediction from subject data

1. A method implemented by a computing device, the method comprising:

collecting, by the computing device, training data that includes word and outcome data;

training, by the computing device, a machine-learning model by tuning a computer representation based on the training data, the training performed to configure the machine-learning model to learn to generate attribution scores based the word and outcome data of the training data;

receiving, by the computing device, subject data that corresponds to digital content, the subject data including a plurality of words;

generating, by the computing device, a plurality of attribution scores using the trained machine-learning model, each attribution score of the plurality of attribution scores describing a respective amount that a respective said word in the subject data contributes toward performance of an outcome;

generating, by the computing device, a plurality of representations based on the plurality of attribution scores, each representation of the plurality of representations including the respective amount that the respective said word in the subject data contributes toward performance of the outcome;

generating, by the computing device, an attribution recommendation based on at least one said attribution score, the attribution recommendation indicating a change to be made to at least one said word to increase a probability of achieving performance of the outcome; and

outputting, by the computing device, the plurality of representations and the attribution recommendation for display in a user interface concurrently with the subject data as indicating the respective amounts that the respective said words in the subject data and the at least one said word contributes toward performance of the outcome.

2. The method as described in claim 1 , wherein the outcome includes receiving a selection of the subject data via a user interface to open the digital content.

3. The method as described in claim 2 , wherein the subject data is a subject line of the digital content, the digital content configured as a digital message.

4. The method as described in claim 2 , wherein the subject data is a title of the digital content.

5. The method as described in claim 1 , wherein the plurality of representations is configured to indicate a positive, negative, or neutral amounts that the respective said word in the subject data contributes toward performance of the outcome.

6. The method as described in claim 1 , wherein the plurality of representations further indicates a respective amount that a phrase including a subject of the plurality of words contributes toward performance of the outcome.

7. The method as described in claim 1 , wherein the plurality of representations further indicates a respective amount that the plurality of words as a whole contributes toward performance of the outcome.

8. The method as described in claim 1 , wherein the attribution recommendation indicates the change using text describing an effect of the at least one word on the achieving of the outcome.

9. The method as described in claim 1 , wherein the generating the plurality of attribution scores is performed by a plurality of different machine-learning models, respectively.

10. The method as described in claim 9 , wherein the plurality of different machine-learning models includes convolutional neural network (CNN) and a long short-term memory (LSTM) neural network.

11. A system comprising:

a training data collection module implemented by a computing device to collect training data that includes word and outcome data;

a model training module implemented by the computing device to train a plurality of machine-learning models based on the training data, the training performed to configure the plurality of machine-learning models to learn to generate outputs as attribution scores that reflect patterns and attributes of the training data on achieving the outcome described by the outcome data;

an input module implemented by the computing device to receive subject data that corresponds to digital content, the subject data including a plurality of words;

an attribute scoring module implemented by the computing device to generate a plurality of attribution scores using a plurality of different types of machine-learning models, respectively, each attribution score of the plurality of attribution scores describing a respective amount that a respective said word in the subject data contributes toward performance of an outcome;

an attribute recommendation module implemented by the computing device to generate an attribution recommendation based on at least one said attribution score, the attribution recommendation indicating a change to be made to at least one said word in the subject data to increase a probability of achieving performance of the outcome and an indication of a corresponding change to the at least one said attribution score based on the change; and

an attribute representation module implemented by the computing device to generate a plurality of representations based on the plurality of attribution scores, each representation of the plurality of representations including the respective amount that the respective said word in the subject data contributes toward performance of the outcome.

12. The system as described in claim 11 , wherein the plurality of different machine-learning models includes convolutional neural network (CNN) and a long short-term memory (LSTM) neural network.

13. The system as described in claim 11 , wherein the outcome includes receiving a selection of the subject data via a user interface to open the digital content.

14. The system as described in claim 11 , wherein the subject data is a subject line of the digital content, the digital content configured as an email.

15. The system as described in claim 11 , wherein the subject data is a title of the digital content.

16. The system as described in claim 11 , wherein the plurality of representations is configured to indicate an amount to which the respective said word in the subject data contributes toward performance of the outcome.

17. The system as described in claim 11 , wherein the plurality of representations further indicates a respective amount that a phrase including a subject of the plurality of words contributes toward performance of the outcome.

18. The system as described in claim 11 , wherein the plurality of representations further indicates a respective amount that the plurality of words as a whole contributes toward performance of the outcome.

19. One or more computer-readable storage media storing instructions that, responsive to execution by a processing system, causes the processing system to perform operations including:

collecting training data that includes word and outcome data;

training a machine-learning model by tuning a computer representation based on the training data, the training performed to configure the machine-learning model to learn to generate attribution scores based the word and outcome data of the training data;

receiving subject data that corresponds to digital content, the subject data including a plurality of words;

generating a plurality of attribution scores using the trained machine-learning model, each attribution score of the plurality of attribution scores describing a respective amount that a respective said word in the subject data contributes toward performance of an outcome;

generating a plurality of representations based on the plurality of attribution scores, each representation of the plurality of representations including the respective amount that the respective said word in the subject data contributes toward performance of the outcome;

generating an attribution recommendation based on at least one said attribution score, the attribution recommendation indicating a change to be made to at least one said word to increase a probability of achieving performance of the outcome; and

outputting the plurality of representations and the attribution recommendation for display in a user interface concurrently with the subject data as indicating the respective amounts that the respective said words in the subject data and the at least one said word contributes toward performance of the outcome.

20. The one or more computer-readable storage media as described in claim 19 , wherein at least one said attribution score is displayable as a gauge in a user interface.