Systems and methods to identify most suitable grammar suggestions among suggestions from a machine translation model

A method for training a grammar accuracy prediction model to determine accuracy of suggested edits to contents of a document, the method comprising:

• generating training data for the grammar accuracy prediction model, wherein generating the training data comprises:
• identifying a sentence that was corrected and a word that was replaced with an edited word in the corrected sentence;
• determining, a semantic dependency relationship and a syntactic dependency relationship between the edited word and each of a set of surrounding words in the corrected sentence, wherein the semantic dependency relationship and the syntactic dependency relationship is determined for at least one of the set of surrounding words irrespective of proximity of a corresponding surrounding word to the edited word;
• generating a first training input comprising the edited word, the edit word and the semantic dependency relationship and the syntactic dependency relationship between the edited word and each of the set of surrounding words;
• and
• generating a first target output for the first training input, wherein the first target output comprises an indication of whether the edited word in the corrected sentence is grammatically accurate in view of the semantic dependency relationship and the syntactic dependency relationship between the edited word and each of the set of surrounding words;
• and
• providing the training data to train the grammar accuracy prediction model on (i) a set of training inputs comprising the first training input and (ii) a set of target outputs comprising the first target output.

A system comprising:

• a memory component;
• and
• a processing device communicatively coupled to the memory component, the processing device to:
• obtain a set of candidate edits for a first word of a sentence, wherein each of the candidate edits comprises an edit word;
• determine, a semantic dependency relationship and a syntactic dependency relationship between the edit word and each of a set of surrounding words in the sentence, wherein the semantic dependency relationship and the syntactic dependency relationship is determined for at least one of the set of surrounding words is identified irrespective of proximity of a corresponding surrounding word to the edit word;
• provide, as input to a grammar accuracy prediction model, the semantic dependency relationship and the syntactic dependency relationship between the edit word and each of the surrounding words and the set of candidate edits;
• obtain one or more first outputs from the grammar accuracy prediction model, wherein the one or more first outputs indicate a grammatical accuracy of each candidate edit from the set in the sentence in view of the semantic dependency relationship and the syntactic dependency relationship between the edit word and each of the set of surrounding words;
• and
• select a candidate edit from the set of candidate edits based on the indicated grammatical accuracy of each candidate edit from the set in the sentence.