Determining structure from a language block

1. An apparatus comprising:

a processor,

a memory that stores code executable by the processor to:

encode phrase of a language block into phrase vectors using a neural network;

determine phrase tags from the phrase vectors,

wherein the phrase tags specify a phrase function;

determine structure tsp for the phrases from the language block;

determine an audience for the language block;

determine an audience preference structure based on the audience;

modify the structure tags based on the audience preference structure and the phrase tags, wherein the audience preference structure is selected from the group consisting of an inductive structure, a deductive structure, a technical structure, a journalistic structure, an artistic structure, a directive structure, and a suggestive structure; and

communicate the phrases based on the modified structure taps.

2. The apparatus of claim 1 , wherein the phrase tags are determined by applying a disambiguation process to the phrase vectors.

3. The apparatus of claim 1 , wherein the code is further executable by the processor to parse the language block into the phrases.

4. The apparatus of claim 1 , wherein the code is further executable by the processor to subtract the structure tags from the phrases.

5. A method comprising:

encoding phrases of a language block into phrase vectors using a neural network;

determining, by use of processor, phrase tags from the phrase vectors, wherein the phrase tags specify a phrase function;

determining structure tags for the phrases from the language block;

determining an audience for the language block;

determining an audience preference structure based on the audience;

modifying the structure tags based on the audience preference structure and the phrase tags, wherein the audience preference structure is selected from the group consisting of an inductive structure, a deductive structure, a technical structure, a journalistic stricture, an artistic structure, a directive structure, and a suggestive structure; and

communicating the phrases based on the modified structure tags.

6. The method of claim 5 , wherein the phrase tags are determined by applying a disambiguation process to the phrase vectors.

7. The method of claim 5 , the method further comprising parsing the language block into the phrases.

8. The method of claim 5 , the method further comprising subtracting the structure tags from the phrases.

9. A program product comprising a computer readable storage medium that stores code executable by a processor, the executable code comprising code to:

encode phrases of a language block into phrase vectors using neural network,

determine phrase tags from the phrase vectors, wherein the phrase tags specify a phrase function;

determine structure tags for the phrases from the language block;

determine an audience preference structure based on the audience;

modify the structure tags based on the audience preference structure and the phrase tags, wherein the audience preference structure is selected from the group consisting of an inductive structure, a deductive structure, a technical structure, a journalistic structure, an artistic structure, a directive structure, and a suggestive structure; and

communicate the phrases based on the modified structure tags.

10. The program product of claim 9 , wherein the phrase tags are determined by applying a disambiguation process to the phrase vectors.