Machine-translation based corrections

A system for identifying viable corrections comprising:

• an interface configured to obtain one or more language snippet pairs, each obtained language snippet pair comprising a first language snippet and a second language snippet;
• a word alignment and scoring module implemented in digital processing logic configured to, for at least one selected language snippet pair of the one or more language snippet pairs, identify the selected language snippet pair as a viable correction pair by:
• computing a word alignment score for an aligned version of the selected language snippet pair;
• and
• determining that the computed word alignment score for the aligned version of the selected language snippet pair is below a word alignment threshold value;
• and
• a character edit distance module implemented in digital processing logic configured to compute a character edit distance for one or more of the aligned words within the aligned version of the selected language snippet pair;
• wherein the word alignment and scoring module is further configured to identify the selected language snippet pair as a viable correction pair by determining that each computed character edit distance for the one or more of the aligned words within the selected language snippet pair is below a character distance threshold value;
• and
• wherein the interface is further configured to provide an indication of the viable correction pair.

A non-transitory computer-readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform operations for identifying viable corrections, the operations comprising:

• obtaining one or more language snippet pairs comprising an initial language snippet and a subsequent language snippet;
• for at least one selected language snippet pair of the one or more language snippet pairs, identifying the selected language snippet pair as a viable correction pair by:
• aligning words between the initial language snippet and the subsequent language snippet of the selected language snippet pair to obtain an aligned version of the selected language snippet pair;
• computing a word alignment score for the aligned version of the selected language snippet pair;
• determining that the computed word alignment score for the aligned version of the selected language snippet pair is below a word alignment threshold value;
• computing a character edit distance for one or more of the aligned words within the aligned version of the selected language snippet pair;
• and
• determining that each computed character edit distance for the one or more of the aligned words within the selected language snippet pair is below a character distance threshold value;
• and
• providing an indication of the viable correction.