--- title: "End-to-end Neural Coreference Resolution — Lee 2017" description: "The first coreference model trained end to end without a parser or a mention detector. Every span up to 10 words is a candidate; each span gets a mention…" authors: ["Kenton Lee", "Luheng He", "Mike Lewis", "Luke Zettlemoyer"] year: 2017 venue: "Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing" source: "https://aclanthology.org/D17-1018/" doi: "10.18653/v1/d17-1018" arxiv: "1707.07045" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/lee-2017-end-end-neural-coreference-resolution/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lee-2017-end-end-neural-coreference-resolution/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2017](https://md.research.elamur.ai/free/en/papers/2017/) / End-to-end Neural Coreference Resolution # End-to-end Neural Coreference Resolution **Authors:** [Kenton Lee](https://md.research.elamur.ai/free/en/authors/kenton-lee/), [Luheng He](https://md.research.elamur.ai/free/en/authors/luheng-he/), [Mike Lewis](https://md.research.elamur.ai/free/en/authors/mike-lewis/), [Luke Zettlemoyer](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) · **Year:** 2017 · **Venue:** EMNLP 2017 **Link:** [aclanthology.org/D17-1018](https://aclanthology.org/D17-1018/) · **Licence:** [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) [**841** citations](https://openalex.org/works?filter=cites:W2738152205) OpenAlex · 2026-10-14 · [**436** citations in our library](https://md.research.elamur.ai/free/en/papers/lee-2017-end-end-neural-coreference-resolution/cited-by/) of the 140673 papers whose bibliographies we read · [Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing](https://aclanthology.org/volumes/D17-1/) · [licence](https://creativecommons.org/licenses/by/4.0/) - [Publisher's page](https://aclanthology.org/D17-1018/) - [PDF](https://aclanthology.org/D17-1018.pdf) - [DOI 10.18653/v1/d17-1018](https://doi.org/10.18653/v1/d17-1018) - [arXiv 1707.07045](https://arxiv.org/abs/1707.07045) - [OpenAlex](https://openalex.org/W2738152205) **Authors (4)** 1. [Kenton Lee](https://openalex.org/A5081862885) [ORCID](https://orcid.org/0000-0002-9534-5970) · [Semantic Scholar](https://www.semanticscholar.org/author/2544107) · h 37 · 52706 citations · 111 works 2. [Luheng He](https://openalex.org/A5025887224) [Semantic Scholar](https://www.semanticscholar.org/author/2265599) · h 25 · 5772 citations · 46 works 3. [Mike Lewis](https://openalex.org/A5004412943) [ORCID](https://orcid.org/0000-0003-0679-6612) · [Semantic Scholar](https://www.semanticscholar.org/author/35084211) · h 38 · 36864 citations · 84 works 4. [Luke Zettlemoyer](https://openalex.org/A5067919401) [ORCID](https://orcid.org/0009-0008-8296-0764) · [Semantic Scholar](https://www.semanticscholar.org/author/1982950) · h 91 · 78722 citations · 492 works Citations and author metrics: OpenAlex, 2026-10-14; authors with the same name are not always told apart. ## What it is about The first coreference model trained end to end without a parser or a mention detector. Every span up to 10 words is a candidate; each span gets a mention score and each pair an antecedent score, and a span chooses among all earlier spans plus a dummy antecedent fixed at score 0 (no link, or not a mention). Span vectors combine BiLSTM states at the boundaries with an attention-weighted «soft head» that learns something close to syntactic heads; pair features are same-speaker, genre, binned distance and span width. Spans are pruned greedily by mention score (0.4 per word, up to 250 antecedents) and the model maximises the marginal likelihood of all gold antecedents. On CoNLL-2012 (OntoNotes) it reaches 67.2 average F1 single and 68.8 as a five-model ensemble, +1.5/+3.1 over the previous best; distance and width features are worth 3.8 F1, and gold mentions would add over 17.5 F1, while a rule-based mention detector costs 1 F1. ## Ideas - I1. Measure how much of `coref`'s error is mention detection: add an oracle-mention switch to `coref eval` (gold mention spans, Mova's sieves for linking) and report CoNLL F1 on GUM dev with gold and SLM trees. The paper's +17.5 F1 for gold mentions sets the expectation; the gap tells whether to work on `coref/src/mention.rs` or on the sieves. - I2. When `coref` moves from sieves to candidate ranking (README «What next»), take this model's decision shape without the network: for each mention, score every earlier candidate with an averaged perceptron over symbolic features (head match, agreement, Hobbs rank, same voice/speaker, the paper's distance buckets 1, 2, 3, 4, 5–7, 8–15, 16–31, 32–63, 64+ in sentences or mentions, mention width), with a dummy «new entity» candidate fixed at 0 so the model abstains when nothing scores above it. Training data must be licence-clean (GUM is NC: dev/test only). The explanation is the feature contributions of the chosen link over the dummy. - I3. Keep the sieve as a precision layer in front: links the speaker, appos and reflexive sieves make at 73–91% precision stay fixed, and the ranker only decides what the sieves leave open. Compare CoNLL F1 on GUM dev for sieve-only, ranker-only and the combination. ## For Mova [languages/en/35-reference-and-coreference.md](https://mova.elamur.ai/free/en/docs/languages/en/35-reference-and-coreference/) §8–9 (algorithms; Mova's coref crate). Code: `coref/src/eval.rs` (oracle mentions), `coref/src/mention.rs`, `coref/src/sieve.rs`, a new ranker next to it. Measures: CoNLL F1 on GUM dev with gold and SLM trees, with and without oracle mentions. Topics: [Coreference resolution](https://md.research.elamur.ai/free/en/topics/coreference-resolution/) ## Cite The paper itself, not this card. BibTeX from the library's record of the work. [DOI 10.18653/v1/d17-1018](https://doi.org/10.18653/v1/d17-1018) · [arXiv 1707.07045](https://arxiv.org/abs/1707.07045) ```bibtex @inproceedings{lee2017end, title = {{End-to-end Neural Coreference Resolution}}, author = {Lee, Kenton and He, Luheng and Lewis, Mike and Zettlemoyer, Luke}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, year = {2017}, doi = {10.18653/v1/d17-1018}, eprint = {1707.07045}, archivePrefix = {arXiv}, url = {https://doi.org/10.18653/v1/d17-1018} } ``` - [← previous Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling](https://md.research.elamur.ai/free/en/papers/marcheggiani-2017-encoding-sentences-graph-convolutional-networks/) - [next → English Multiword Expression-aware Dependency Parsing Including Named Entities](https://md.research.elamur.ai/free/en/papers/aclbulk-2017-kato-english-multiword-expression-aware-dependency/) - [↑ back to papers of 2017](https://md.research.elamur.ai/free/en/papers/2017/) ## References 1 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Martın Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, et al. 2015. TensorFlow: Large-scale Machine Learning on Heterogeneous Systems. Software available from tensorflow.org. [find](https://scholar.google.com/scholar?q=Mart%C4%B1n+Abadi%2C+Ashish+Agarwal%2C+Paul+Barham%2C+Eugene+Brevdo%2C+Zhifeng+Chen%2C+Craig+Citro%2C+Greg+S+Corrado%2C+Andy+Davis%2C+Jeffrey+Dean%2C+Matthieu+Devin%2C+et+al%2E+2015%2E+TensorFlow%3A+Large%2Dscale+Machine+Learning+on+)