Automated Grammatical Error Correction (GEC) Task
(Redirected from GEC)
Jump to navigation
Jump to search
An Automated Grammatical Error Correction (GEC) Task is automated linguistic error correction task that corrects grammatical errors.
- Context:
- It can be solved by a Grammatical Error Correction System (that implements a grammatical error correction algorithm).
- It can be supported by a Grammatical Error Detection Task.
- It can range from being a Phrase-input GEC Task, Sentence-input GEC Task, Passage-input GEC Task, Document-input GEC Task, ...
- …
- Example(s):
- GEC(“
I like yesterdays weather.
”) ⇒ “I liked yesterday's weather.
". - GEC(“
I'm going to store tomorow.
”) ⇒ “I'm going to the store tomorow.
". - CoNLL-2014 Shared Task on Grammatical Error Correction.
- …
- GEC(“
- Counter-Example(s):
- See: Text Sentence, Transcription Error.
References
2018
- (Ge et al., 2018) ⇒ Tao Ge, Furu Wei, and Ming Zhou. (2018). “Reaching Human-level Performance in Automatic Grammatical Error Correction: An Empirical Study.” arXiv preprint arXiv:1807.01270
- ABSTRACT: Neural sequence-to-sequence (seq2seq) approaches have proven to be successful in grammatical error correction (GEC). Based on the seq2seq framework, we propose a novel fluency boost learning and inference mechanism. Fluency boosting learning generates diverse error-corrected sentence pairs during training, enabling the error correction model to learn how to improve a sentence's fluency from more instances, while fluency boosting inference allows the model to correct a sentence incrementally with multiple inference steps. Combining fluency boost learning and inference with convolutional seq2seq models, our approach achieves the state-of-the-art performance: 75.72 (F_{0.5}) on CoNLL-2014 10 annotation dataset and 62.42 (GLEU) on JFLEG test set respectively, becoming the first GEC system that reaches human-level performance (72.58 for CoNLL and 62.37 for JFLEG) on both of the benchmarks.
2014
- (Ng et al., 2014) ⇒ Hwee Tou Ng, Siew Mei Wu, Ted Briscoe, Christian Hadiwinoto, Raymond Hendy Susanto, and Christopher Bryant. (2014). “The CoNLL-2014 Shared Task on Grammatical Error Correction.” In: Proceedings of the Eighteenth Conference on Computational Natural Language Learning: Shared Task.
- QUOTE: ... Grammatical error correction is the shared task of the Eighteenth Conference on Computational Natural Language Learning in 2014 (CoNLL-2014). In this task, given an English essay written by a learner of English as a second language, the goal is to detect and correct the grammatical errors of all error types present in the essay, and return the corrected essay. ...
... Grammatical errors consist of many different types, including articles or determiners, prepositions, noun form, verb form, subject-verb agreement, pronouns, word choice, sentence structure, punctuation, capitalization, etc. However, most prior published research on grammatical error correction only focuses on a small number of frequently occurring error types, such as article and preposition errors (Han et al., 2006; Gamon, 2010; Rozovskaya and Roth, 2010; Tetreault et al., 2010; Dahlmeier and Ng, 2011b).
- QUOTE: ... Grammatical error correction is the shared task of the Eighteenth Conference on Computational Natural Language Learning in 2014 (CoNLL-2014). In this task, given an English essay written by a learner of English as a second language, the goal is to detect and correct the grammatical errors of all error types present in the essay, and return the corrected essay. ...
Type | Description | Example |
---|---|---|
Vt | Verb tense | Medical technology during that time [is → was] not advanced enough to cure him. |
Vm | Verb modal | Although the problem [would → may] not be serious, people [would → might] still be afraid. |
V0 | Missing verb | However, there are also a great number of people [who → who are] against this technology. |
Vform | Verb form | A study in 2010 [shown → showed that patients recover faster when surrounded by family members. |
SVA | Subject-verb agreement | The benefits of disclosing genetic risk information [outweighs → outweigh] the costs. |
ArtOrDet | Article or determiner | It is obvious to see that [internet → the internet] saves people time and also connects people globally. |
Nn | Noun number | A carrier may consider not having any [child → children] after getting married. |
Npos | Noun possessive | Someone should tell the [carriers → carrier’s] relatives about the genetic problem. |
Pform | Pronoun form | A couple should run a few tests to see if [their → they] have any genetic diseases beforehand. |
Pref | Pronoun reference | It is everyone’s duty to ensure that [he or she → they] undergo regular health checks. |
Prep | Preposition | This essay will [discuss about → discuss] whether a carrier should tell his relatives or not. |
Wci | Wrong collocation idiom | Early examination is [healthy → advisable] and will cast away unwanted doubts. |
Wa | Acronyms | After [WOWII → World War II], the population of China decreased rapidly. |
Wform | Word form | The sense of [guilty → guilt] can be more than expected. |
Wtone | Tone (formal/informal) | [It’s → It is] our family and relatives that bring us up. |
Srun | Run-on sentences, comma splices | The issue is highly [debatable, a → debatable. A] genetic risk could come from either side of the family. |
Smod | Dangling modifiers | [Undeniable, → It is undeniable that] it becomes addictive when we spend more time socializing virtually. |
Spar | Parallelism | We must pay attention to this information and [assisting → assist] those who are at risk. |
Sfrag | Sentence fragment | However, from the ethical point of view. |
Ssub | Subordinate clause | This is an issue [needs → that needs] to be addressed. |
WOinc | Incorrect word order | [Someone having what kind of disease → What kind of disease someone has] is a matter of their own privacy. |
WOadv | Incorrect adjective/ adverb order | In conclusion, [personally I → I personally] feel that it is important to tell one’s family members. |
Trans | Linking words/phrases | It is sometimes hard to find [out → out if] one has this disease. |
Mec | Spelling, punctuation, capitalization, etc. | This knowledge [maybe relavant → may be relevant] to them. |
Rloc− | Redundancy | It is up to the [patient’s own choice → patient] to disclose information. |
Cit | Citation | Poor citation practice. |
Others | Other errors | An error that does not fit into any other category but can still be corrected. |
Um | Unclear meaning | Genetic disease has a close relationship with the born gene. (i.e., no correction possible without further clarification.) |