sebastianruder--nlp-progress
Mehrdad Farahani
f3f86e3eec
* First version of summarization for persian * First version of summarization for persian docs * Fix misspell, add NER section, update ARMAN table * Update PEYMA table * Add NLI information for persian * Add new Vietnamese POS tagging results (#508) * Add new Vietnamese POS tagging results * Update vietnamese.md * Fix dataset title in semantic_parsing.md (#519) LDC2016E25 is the release limited to DEFT participants in 2016. LDC2017T10 is the general public release of this data and the name used in all recent papers. I believe the data is exactly the same (it definitely has the same number of train/test/dev samples). * Add New Result for Constituency Parsing on PTB (#516) Strongly Incremental Constituency Parsing with Graph Neural Networks Kaiyu Yang and Jia Deng Neural Information Processing Systems (NeurIPS) 2020 * add logical reasoning dataset ReClor (#515) * Update AMR parsing results (#514) I would like to "Create a new branch for this commit and start a pull request" * Added DRS parsing results for PMB 2.2.0 and 3.0.0 (#510) * add a new summarization result (#505) * add new update to Simplification (#504) * add new update to Simplification * Update english/simplification.md Co-authored-by: HaukurPall <haukurpalljonsson@gmail.com> Co-authored-by: Sebastian Ruder <sebastian@ruder.io> Co-authored-by: HaukurPall <haukurpalljonsson@gmail.com> * Add POS tagging result - social media (#503) * Add POS tagging result - social media * Apply suggestions from code review * Update english/part-of-speech_tagging.md Co-authored-by: Sebastian Ruder <sebastian@ruder.io> * Add a new task, NLI for Persian (#501) * Add a new task, NLI for Persian * Rename natural-language-inference to natural-language-inference.md Co-authored-by: Sebastian Ruder <sebastian@ruder.io> * Added joint entity and relation extraction task in relationship_extraction.md (#473) * Update relationship_extraction.md * Update relationship_extraction.md * Update relationship_extraction.md * Update semantic_parsing.md (#522) * Update semantic_parsing.md * Update semantic_parsing.md * Add dialogue implementations (#521) * Add SGNN implementation Add link to SGNN implementation * Add CASA implementation Add link to CASA implementation * Update word_sense_disambiguation.md (#517) * Update semantic_parsing.md (#518) Recently released AMR parser using a Seq-to-Seq model. Code can be found at... https://github.com/xdqkid/S2S-AMR-Parser Co-authored-by: Sebastian Ruder <ruder.sebastian@gmail.com> * First version of summarization for persian * First version of summarization for persian docs * Fix misspell, add NER section, update ARMAN table * Update PEYMA table * Add NLI information for persian * Rename nli, fix nli-results, fix typos in readme Co-authored-by: Huy Ngo <mayngulama1@gmail.com> Co-authored-by: Brad Jascob <bjascob@msn.com> Co-authored-by: Kaiyu Yang <kaiyuy@princeton.edu> Co-authored-by: Weihao Yu <1090924009@qq.com> Co-authored-by: Young-Suk Lee <ysuklee@us.ibm.com> Co-authored-by: Rik van Noord <31345314+RikVN@users.noreply.github.com> Co-authored-by: douglas01996 <437527550@qq.com> Co-authored-by: Billie <chuang211@163.com> Co-authored-by: Sebastian Ruder <sebastian@ruder.io> Co-authored-by: HaukurPall <haukurpalljonsson@gmail.com> Co-authored-by: Sara <59834747+sara-meftah@users.noreply.github.com> Co-authored-by: h-amirkhani <h.amirkhani.n@gmail.com> Co-authored-by: Tapas <tnk02.05@gmail.com> Co-authored-by: Andres Suarez <suarezandres@gmail.com> Co-authored-by: Gabor Berend <berendg@inf.u-szeged.hu> Co-authored-by: Sebastian Ruder <ruder.sebastian@gmail.com>
2.9 KiB
2.9 KiB
Named entity recognition
Named entity recognition (NER) is the task of tagging entities in text with their corresponding type. Approaches typically use BIO notation, which differentiates the beginning (B) and the inside (I) of entities. O is used for non-entity tokens.
Example:
| Mark | Watney | visited | Mars |
|---|---|---|---|
| B-PER | I-PER | O | B-LOC |
ArmanPersoNERCorpus
The ArmanPersoNERCorpus dataset contains 7,682 sentences with 250,015 tokens tagged in IOB format in six different classes, Organization, Person, Location, Facility, Event, and Product.
Download Links: ARMAN
| Model | F1 | Paper / Source | Code |
|---|---|---|---|
| ParsBERT (Farahani et al., 2020) | 99.84 | ParsBERT: Transformer-based Model for Persian Language Understanding | Official |
| LSTM-CRF (Hafezi, Rezaeian, 2018) | 86.55 | Neural Architecture for Persian Named Entity Recognition | - |
| mBERT (Taher et al., 2020) | 84.03 | Beheshti-NER: Persian Named Entity Recognition Using BERT | Official |
| Deep-CRF (Bokaei, Mahmoudi, 2018) | 81.50 | Improved Deep Persian Named Entity Recognition | - |
| Deep-Local (Bokaei, Mahmoudi, 2018) | 79.19 | Improved Deep Persian Named Entity Recognition | - |
| BiLSTM-CRF (Poostchi et al., 2018) | 77.45 | BiLSTM-CRF for Persian Named-Entity Recognition | - |
| SVM-HMM (Poostchi et al., 2016) | 72.59 | PersoNER: Persian Named-Entity Recognition | - |
PEYMA
The PEYMA dataset includes 7,145 sentences with 302,530 tokens from which 41,148 tokens are tagged in IOB format in with seven different classes, Organization, Percent, Money, Location, Date, Time, and Person.
Download Links: PEYMA
| Model | F1 | Paper / Source | Code |
|---|---|---|---|
| ParsBERT (Farahani et al., 2020) | 93.40 | ParsBERT: Transformer-based Model for Persian Language Understanding | Official |
| mBERT (Taher et al., 2020) | 90.59 | Beheshti-NER: Persian Named Entity Recognition Using BERT | Official |
| Rule-Based-CRF (Shahshahani et al., 2018) | 84.00 | PEYMA: A Tagged Corpus for Persian Named Entities | - |