Text correction for dyslexic users on an online social network
In one embodiment, a method includes identifying a plurality of dyslexic users on an online social network. The plurality of dyslexic users may be identified based on content objects posted by these users over a particular time period, where the content objects may include one or more of word-level errors or sentence-level errors. A machine-learning model may be trained for text correction using a corpus of social network data, which may include at least the content objects with one or more of word-level errors or sentence-level errors, and a corresponding set of corrected content objects. A text string including one or more errors may be received from a client system associated with a first user. The text string may be transformed into a vector representation using an encoder of the machine-learning model. A corrected text string may be generated from the vector representation using a decoder of the machine-learning model.
1. A method comprising, by one or more computing systems:
identifying a plurality of dyslexic users on an online social network, wherein the plurality of dyslexic users are identified based on a set of content objects posted by the dyslexic users over a particular time period, the content objects posted by the dyslexic users comprising one or more of word-level errors or sentence-level errors;
training a machine-learning model for text correction using a corpus of social network data, the social network data comprising at least the set of content objects posted by the dyslexic users with one or more of word-level errors or sentence-level errors, and a corresponding set of corrected content objects that are posted to replace the posted set of content objects;
receiving, from a client system associated with a first user of an online social network, a text string, the text string comprising one or more errors;
transforming, using an encoder of the machine-learning model, the text string into a vector representation;
generating, using a decoder of the machine-learning model, a corrected text string from the vector representation, wherein the corrected text string has the one or more errors removed; and
sending, to the client system associated with the first user, instructions for presenting the corrected text string.
2. The method of claim 1 , wherein the set of content objects posted by the dyslexic users comprises social media posts.
3. The method of claim 2 , wherein:
a content object, posted by a dyslexic user from the plurality of dyslexic users, with one or more of word-level errors or sentence-level errors comprises a first social media post; and
a corresponding corrected content object comprises a second social media post submitted by the dyslexic user to correct the first social media post.
4. The method of claim 1 , further comprising: determining that the first user is dyslexic.
5. The method of claim 1 , wherein the word-level errors comprises one or more of:
words with swapping letters;
words with omitting letters;
foreshortening words;
words omitting prefixes or suffixes; or
words with capitalization issues.
6. The method of claim 1 , wherein the sentence-level errors comprises one or more of:
grammar errors;
wrong prepositions;
odd or no spacing between words;
sentence fragment issues; or
wrong punctuations.
7. The method of claim 1 , further comprising:
receiving, from the client system associated with the first user of the online social network, a request to post the text string on the online social network, wherein the corrected text string is presented on the client system as a suggestion to the first user in response to receiving the request.
8. The method of claim 7 , further comprising:
receiving, from the client system associated with the first user of the online social network, a confirmation from the first user accepting the corrected text string; and
posting the corrected text string on the online social network.
9. The method of claim 1 , further comprising:
generating, using the machine-learning model, suggestions for correcting the one or more errors in the text string in real-time.
10. The method of claim 1 , wherein transforming the text string into the vector representation preserves the semantic meaning of the text string.
11. The method of claim 1 , further comprising:
in response to receiving the text string, analyzing the text; and
categorizing the text, based on the analysis, into one of a non-English text, a well-written text, an understandable but informal text, and a poorly-written text.
12. The method of claim 11 , further comprising:
determining that the text is categorized into the poorly-written text.
13. The method of claim 1 , wherein the machine-learning model is a sequence-to-sequence model.
14. The method of claim 1 , further comprising:
parsing the text string into one or more n-grams, wherein the one or more n-grams comprises the one or more errors.
15. The method of claim 1 , wherein the one or more n-grams are one or more of unigrams, bigrams, or trigrams.
16. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:
identify a plurality of dyslexic users on an online social network, wherein the plurality of dyslexic users are identified based on a set of content objects posted by the dyslexic users over a particular time period, the content objects posted by the dyslexic users comprising one or more of word-level errors or sentence-level errors;
train a machine-learning model for text correction using a corpus of social network data, the social network data comprising at least the set of content objects posted by the dyslexic users with one or more of word-level errors or sentence-level errors, and a corresponding set of corrected content objects that are posted to replace the posted set of content objects;
receive, from a client system associated with a first user of an online social network, a text string, the text string comprising one or more errors;
transform, using an encoder of the machine-learning model, the text string into a vector representation;
generate, using a decoder of the machine-learning model, a corrected text string from the vector representation, wherein the corrected text string has the one or more errors removed; and
send, to the client system associated with the first user, instructions for presenting the corrected text string.
17. The media of claim 16 , wherein the set of content objects posted by the dyslexic users comprises social media posts.
18. The media of claim 17 , wherein:
a content object, posted by a dyslexic user from the plurality of dyslexic users, with one or more of word-level errors or sentence-level errors comprises a first social media post; and
a corresponding corrected content object comprises a second social media post submitted by the dyslexic user to correct the first social media post.
19. A system comprising: one or more processors; and a non-transitory memory coupled to the processors comprising instructions executable by the processors, the processors operable when executing the instructions to:
identify a plurality of dyslexic users on an online social network, wherein the plurality of dyslexic users are identified based on a set of content objects posted by the dyslexic users over a particular time period, the content objects posted by the dyslexic users comprising one or more of word-level errors or sentence-level errors;
train a machine-learning model for text correction using a corpus of social network data, the social network data comprising at least the set of content objects posted by the dyslexic users with one or more of word-level errors or sentence-level errors, and a corresponding set of corrected content objects that are posted to replace the posted set of content objects;
receive, from a client system associated with a first user of an online social network, a text string, the text string comprising one or more errors;
transform, using an encoder of the machine-learning model, the text string into a vector representation;
generate, using a decoder of the machine-learning model, a corrected text string from the vector representation, wherein the corrected text string has the one or more errors removed; and
send, to the client system associated with the first user, instructions for presenting the corrected text string.
20. The media of claim 19 , wherein the set of content objects posted by the dyslexic users comprises social media posts.