---
title: "Leveraging Label Semantics and Entity Description Generation"
description: "The authors propose DR-FET, which uses short natural-language descriptors as a bridge between mentions and labels. 19 references."
authors: ["Yingying Ma", "Hongliang Dai", "Xia Zhang", "Piji Li"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.1584/"
doi: "10.18653/v1/2026.findings-acl.1584"
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/ma-2026-leveraging-label-semantics-entity-description/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ma-2026-leveraging-label-semantics-entity-description/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/34/) / Leveraging Label Semantics and Entity Description Generation for LLM-based Fine-grained Entity Typing

# Leveraging Label Semantics and Entity Description Generation for LLM-based Fine-grained Entity Typing

**Authors:** Yingying Ma, Hongliang Dai, Xia Zhang and [Piji Li](https://md.research.elamur.ai/free/en/authors/piji-li/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.1584/) · [PDF](https://aclanthology.org/2026.findings-acl.1584.pdf) · [DOI 10.18653/v1/2026.findings-acl.1584](https://doi.org/10.18653/v1/2026.findings-acl.1584) · [OpenAlex](https://openalex.org/W7166828836)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Fine-grained entity typing (FET) assigns detailed, context-dependent types to entity mentions. LLMs struggle with it for two reasons: ontologies contain dozens to hundreds of types, and training data from distant supervision is noisy. The authors propose DR-FET, which uses short natural-language descriptors as a bridge between mentions and labels.

An LLM first writes several descriptors for each mention. Each type is represented by the descriptors that occur most often among its training instances. A frozen text encoder then matches entity descriptors against label descriptors and keeps a small set of likely types. A LoRA-tuned open-source LLM makes the final multi-label choice from that shortlist only.

The same retrieval step also cleans the training data. It keeps only distantly labelled examples whose retrieved candidates agree with the distant labels, which yields a small, high-precision fine-tuning set without any manual annotation.

On two standard benchmarks, DR-FET beats earlier FET approaches under noisy supervision. This matters because it turns a large-label-space problem into a constrained decision, an idea that could carry over to other weakly supervised classification tasks.

## Key points
- Motivating experiment: a strong LLM reaches 65.61% strict accuracy when picking among 5 candidate types, but only 29.4% when facing 50 types.
- The pipeline has three stages: descriptor generation (3–5 per mention, few-shot), retrieval scored by summed cosine similarity, and candidate-constrained inference with a LoRA-tuned LLM. The number of label descriptors is fixed at Q = 5.
- Retrieval tests compared the BERT, BGE and E5 encoders. Descriptor-based retrieval exceeded 95% Recall@K with fewer than seven candidates, while a BERT+Cls logit-ranking baseline fell clearly behind.
- Results hold across backbones: Qwen2.5-3B-Instruct, Gemma-2-2B-it and LLaMA-2-7B-chat-hf reach roughly 69.7–72.6% strict accuracy, so the method does not depend on one specific LLM.
- Code is released at github.com/BlazingOut/DR-FET. The authors list three limitations: descriptor quality varies with the LLM, recall is not guaranteed for rare types, and the multi-stage pipeline adds computational cost.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Nanjing University of Aeronautics and Astronautics](https://md.research.elamur.ai/free/en/universities/nanjing-university-of-aeronautics-and-astronautics/) — University, Nanjing, China · 4 works · 2 citations

Also on the byline: [Nanjing University of Chinese Medicine](https://ror.org/04523zj19).

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works

🧭 Topics (1): [🏷️ Named entity recognition](https://md.research.elamur.ai/free/en/topics/named-entity-recognition/) · 200 works

## Cite

Cite as: Yingying Ma, Hongliang Dai, Xia Zhang and Piji Li. 2026. Leveraging Label Semantics and Entity Description Generation for LLM-based Fine-grained Entity Typing. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{ma2026leveraging,
  title = {{Leveraging Label Semantics and Entity Description Generation for LLM-based Fine-grained Entity Typing}},
  author = {Yingying Ma, Hongliang Dai, Xia Zhang and Piji Li},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1584},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1584}
}
```

- [← previous Leveraging Digitized Newspapers to Collect Summarization Data in Low-Resource Languages](https://md.research.elamur.ai/free/en/papers/dahan-2026-leveraging-digitized-newspapers-collect-summarization/)
- [next → Leveraging Outline-Optimized Generative Interactions and Critique for Self-Refining Outlines with Reinforcement Learning](https://md.research.elamur.ai/free/en/papers/liu-2026-leveraging-outline-optimized-generative-interactions/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/34/)

## References

19 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Bo Chen, Xiaotao Gu, Yufeng Hu, Siliang Tang, Guoping Hu, Yueting Zhuang, and Xiang Ren. 2019. Improving distantly-supervised entity typing with compact latent space clustering. arXiv preprint. [find](https://scholar.google.com/scholar?q=Bo+Chen%2C+Xiaotao+Gu%2C+Yufeng+Hu%2C+Siliang+Tang%2C+Guoping+Hu%2C+Yueting+Zhuang%2C+and+Xiang+Ren%2E+2019%2E+Improving+distantly%2Dsupervised+entity+typing+with+compact+latent+space+clustering%2E+arXiv+preprint%2E)
2. Gemma Team. 2024. Gemma. Accessed: 2024-12-26. [find](https://scholar.google.com/scholar?q=Gemma+Team%2E+2024%2E+Gemma%2E+Accessed%3A+2024%2D12%2D26%2E)
3. Dan Gillick, Nevena Lazic, Kuzman Ganchev, Jesse Kirchner, and David Huynh. 2016. Context-dependent fine-grained entity type tagging. arXiv preprint. [find](https://scholar.google.com/scholar?q=Dan+Gillick%2C+Nevena+Lazic%2C+Kuzman+Ganchev%2C+Jesse+Kirchner%2C+and+David+Huynh%2E+2016%2E+Context%2Ddependent+fine%2Dgrained+entity+type+tagging%2E+arXiv+preprint%2E)
4. Jui-Ting Huang, Ashish Sharma, Shuying Sun, Li Xia, David Zhang, Philip Pronin, Janani Padmanabhan, Giuseppe Ottaviano, and Linjun Yang. 2020. Embedding-based retrieval in facebook search. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 2553… [find](https://scholar.google.com/scholar?q=Jui%2DTing+Huang%2C+Ashish+Sharma%2C+Shuying+Sun%2C+Li+Xia%2C+David+Zhang%2C+Philip+Pronin%2C+Janani+Padmanabhan%2C+Giuseppe+Ottaviano%2C+and+Linjun+Yang%2E+2020%2E+Embedding%2Dbased+retrieval+in+facebook+search%2E+In+Proceedi)
5. Tanay Komarlu, Minhao Jiang, Xuan Wang, and Jiawei Han. 2024. OntoType: Ontology-guided and pre-trained language model assisted fine-grained entity typing. arXiv preprint. [find](https://scholar.google.com/scholar?q=Tanay+Komarlu%2C+Minhao+Jiang%2C+Xuan+Wang%2C+and+Jiawei+Han%2E+2024%2E+OntoType%3A+Ontology%2Dguided+and+pre%2Dtrained+language+model+assisted+fine%2Dgrained+entity+typing%2E+arXiv+preprint%2E)
6. Bangzheng Li, Wenpeng Yin, and Muhao Chen. 2022. Ultra-fine entity typing with indirect supervision from natural language inference. Transactions of the Association for Computational Linguistics, 10:607– 622. [find](https://scholar.google.com/scholar?q=Bangzheng+Li%2C+Wenpeng+Yin%2C+and+Muhao+Chen%2E+2022%2E+Ultra%2Dfine+entity+typing+with+indirect+supervision+from+natural+language+inference%2E+Transactions+of+the+Association+for+Computational+Linguistics%2C+10%3A6)
7. Frank Mtumbuka and Steven Schockaert. 2024. EnCore: Fine-grained entity typing by pre-training entity encoders on coreference chains. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1768–1781, St. Julian’s,… [find](https://scholar.google.com/scholar?q=Frank+Mtumbuka+and+Steven+Schockaert%2E+2024%2E+EnCore%3A+Fine%2Dgrained+entity+typing+by+pre%2Dtraining+entity+encoders+on+coreference+chains%2E+In+Proceedings+of+the+18th+Conference+of+the+European+Chapter+of+t)
8. Yasumasa Onoe and Greg Durrett. 2019. Learning to denoise distantly-labeled data for entity typing. arXiv preprint. [find](https://scholar.google.com/scholar?q=Yasumasa+Onoe+and+Greg+Durrett%2E+2019%2E+Learning+to+denoise+distantly%2Dlabeled+data+for+entity+typing%2E+arXiv+preprint%2E)
9. Weiran Pan, Wei Wei, and Feida Zhu. 2022. Automatic noisy label correction for fine-grained entity typing. arXiv preprint. [find](https://scholar.google.com/scholar?q=Weiran+Pan%2C+Wei+Wei%2C+and+Feida+Zhu%2E+2022%2E+Automatic+noisy+label+correction+for+fine%2Dgrained+entity+typing%2E+arXiv+preprint%2E)
10. Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, Rohan Anil, Zakaria Haque, Lichan Hong, Vihan Jain, Xiaobing Liu, and Hemal Shah. 2016. Wide & deep learning for recommender systems. In Proceedings of the… [find](https://scholar.google.com/scholar?q=Heng%2DTze+Cheng%2C+Levent+Koc%2C+Jeremiah+Harmsen%2C+Tal+Shaked%2C+Tushar+Chandra%2C+Hrishi+Aradhye%2C+Glen+Anderson%2C+Greg+Corrado%2C+Wei+Chai%2C+Mustafa+Ispir%2C+Rohan+Anil%2C+Zakaria+Haque%2C+Lichan+Hong%2C+Vihan+Jain%2C+Xiao)
11. Kunyuan Pang, Haoyu Zhang, Jie Zhou, and Ting Wang. 2022. Divide and denoise: Learning from noisy labels in fine-grained entity typing with cluster-wise loss correction. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1997–200… [find](https://scholar.google.com/scholar?q=Kunyuan+Pang%2C+Haoyu+Zhang%2C+Jie+Zhou%2C+and+Ting+Wang%2E+2022%2E+Divide+and+denoise%3A+Learning+from+noisy+labels+in+fine%2Dgrained+entity+typing+with+cluster%2Dwise+loss+correction%2E+In+Proceedings+of+the+60th+Ann)
12. Hongliang Dai and Ziqian Zeng. 2023. From ultra-fine to fine: Fine-tuning ultra-fine entity typing models to fine-grained. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2259–2270, Toronto, Canada. Association for Cmputationa… [find](https://scholar.google.com/scholar?q=Hongliang+Dai+and+Ziqian+Zeng%2E+2023%2E+From+ultra%2Dfine+to+fine%3A+Fine%2Dtuning+ultra%2Dfine+entity+typing+models+to+fine%2Dgrained%2E+In+Proceedings+of+the+61st+Annual+Meeting+of+the+Association+for+Computationa)
13. Qwen Team. 2024. Qwen2.5: A party of foundation models. [find](https://scholar.google.com/scholar?q=Qwen+Team%2E+2024%2E+Qwen2%2E5%3A+A+party+of+foundation+models%2E)
14. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. [BERT: Pre-training of deep bidirectional transformers for language understanding](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/). arXiv preprint.
15. Haochen Shi, Siliang Tang, Xiaotao Gu, Bo Chen, Zhigang Chen, Jian Shao, and Xiang Ren. 2020. Alleviate dataset shift problem in fine-grained entity typing with virtual adversarial training. In Proceedings of the 29th International Joint Conference on Artificial Intelligence, pages 3898–3904, Yokoha… [find](https://scholar.google.com/scholar?q=Haochen+Shi%2C+Siliang+Tang%2C+Xiaotao+Gu%2C+Bo+Chen%2C+Zhigang+Chen%2C+Jian+Shao%2C+and+Xiang+Ren%2E+2020%2E+Alleviate+dataset+shift+problem+in+fine%2Dgrained+entity+typing+with+virtual+adversarial+training%2E+In+Procee)
16. Alejandro Sierra-Múnera, Jan Westphal, and Ralf Krestel. 2023. Efficient ultrafine typing of named entities. In Proceedings of the 2023 ACM/IEEE Joint Conference on Digital Libraries (JCDL), pages 205–214, Santa Fe, New Mexico, USA. IEEE. [find](https://scholar.google.com/scholar?q=Alejandro+Sierra%2DM%C3%BAnera%2C+Jan+Westphal%2C+and+Ralf+Krestel%2E+2023%2E+Efficient+ultrafine+typing+of+named+entities%2E+In+Proceedings+of+the+2023+ACM%2FIEEE+Joint+Conference+on+Digital+Libraries+%28JCDL%29%2C+pages+205)
17. Minghao Tang, Yongquan He, Yongxiu Xu, Hongbo Xu, Wenyuan Zhang, and Yang Lin. 2023. Learning to correct noisy labels for fine-grained entity typing via co-prediction prompt tuning. In Findings of the Association for Computational Linguistics: EMNLP, pages 5848–5858, Singapore. Association for Compu… [find](https://scholar.google.com/scholar?q=Minghao+Tang%2C+Yongquan+He%2C+Yongxiu+Xu%2C+Hongbo+Xu%2C+Wenyuan+Zhang%2C+and+Yang+Lin%2E+2023%2E+Learning+to+correct+noisy+labels+for+fine%2Dgrained+entity+typing+via+co%2Dprediction+prompt+tuning%2E+In+Findings+of+the)
18. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, and 49 others… — [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://md.research.elamur.ai/free/en/papers/touvron-2023-llama-2-open-foundation-fine/)
19. Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. [Text embeddings by weakly-supervised contrastive pre-training](https://md.research.elamur.ai/free/en/papers/wang-2022-text-embeddings-weakly-supervised-contrastive/). arXiv preprint.

*Updated 1 November 2026*