---
title: "The Role of Common Ground for Referential Expressions in"
description: "Coreference systems are usually developed on corpora like Ontonotes, where documents are… 20 references, cited by 3 papers here."
authors: ["Jaap Kruijt", "Piek Vossen"]
year: 2022
venue: "Proceedings of the Fifth Workshop on Computational Models of Reference, Anaphora and Coreference"
source: "https://aclanthology.org/2022.crac-1.10/"
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/kruijt-2022-role-common-ground-referential-expressions/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/kruijt-2022-role-common-ground-referential-expressions/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2022](https://md.research.elamur.ai/free/en/papers/2022/page/6/) / The Role of Common Ground for Referential Expressions in Social Dialogues

# The Role of Common Ground for Referential Expressions in Social Dialogues

**Authors:** Jaap Kruijt and [Piek Vossen](https://md.research.elamur.ai/free/en/authors/piek-vossen/) · **Year:** 2022 · **Venue:** [Proceedings of the Fifth Workshop on Computational Models of Reference, Anaphora and Coreference](https://md.research.elamur.ai/free/en/venues/crac/) ([CRAC 2022](https://md.research.elamur.ai/free/en/venues/crac/2022/)) · **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/2022.crac-1.10/) · [PDF](https://aclanthology.org/2022.crac-1.10.pdf) · [OpenAlex](https://openalex.org/W6948426772)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [3](https://md.research.elamur.ai/free/en/papers/kruijt-2022-role-common-ground-referential-expressions/cited-by/) in our library, of the 141,840 papers whose bibliographies we read

## Summary
Coreference systems are usually developed on corpora like Ontonotes, where documents are unrelated to each other. As a result, they never face the way close friends talk about people they all know. The authors treat third-person reference in social dialogue as something that changes as shared knowledge accumulates. They split referents into an inner circle (people in the speakers' common ground) and an outer circle (people who matter only locally). They reorganise the SemEval 2018 task 4 FRIENDS transcripts, which carry gold person mentions in CoNLL-2012 format, into a chronologically ordered coreference task. They then evaluate SpanBERT off the shelf and after fine-tuning on growing amounts of earlier episodes. The analysis shows that inner and outer circle people are named and introduced differently. Test episodes with a larger share of inner circle mentions get lower scores. Fine-tuning on earlier episodes reduces missed links but adds spurious ones, which points to overfitting rather than real acquisition of common ground. The authors conclude that dialogue agents need more explicit, knowledge-rich modeling of shared history.

## Key points
- Data: SemEval 2018 task 4 (character identification) on FRIENDS seasons 1–2, in CoNLL-2012 format, with gold mentions restricted to people. Its list of 401 characters was split into inner and outer circle, with the six main characters always inner.
- Model: SpanBERT via the Xu and Choi (2020) implementation, tested pretrained and as three fine-tuned variants (small, medium, large portions of strictly preceding episodes).
- Test episodes: S2E14, with an inner/outer ratio of about 4/1, was harder for every model than S2E24, at about 1/1. On most metrics the pretrained model did best on S2E14 and the large fine-tuned model did best on S2E24. On entity-based metrics the pretrained model led on both.
- Inner circle people tend to be mentioned by name or with short kin terms such as "mom". Outer circle people get fuller introductions followed by pronoun chains.
- Code is released at github.com/cltl/inner-outer-coreference.

## Interesting ideas
- Fine-tune only on material that comes chronologically before the test portion, so that "more context" really simulates accumulated history instead of leaking future information.
- Label each referent by how established it is in the shared background (recurring cast vs one-off figure) and report coreference errors separately for each group.
- Pick test sets whose mix of familiar and unfamiliar referents differs on purpose (4/1 vs 1/1), so the comparison exposes sensitivity to one variable.
- Track false positives and false negatives separately, broken down by POS and referent class, since an aggregate score can hide gains in recall that are paid for with overfitted spurious links.
- A nickname chain was split into two clusters because another person was mentioned in between and the sliding window dropped the earlier mentions. Entity memory should therefore not depend on a local window.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (learn): when Mova adapts to a long text or series of tales, train or induce rules only from earlier chapters and evaluate on later ones, to measure genuine accumulation.
- **I2** (frames): split coreference evaluation into recurring protagonists and one-scene characters in fables, to see whether failures concentrate on familiar entities introduced by bare names or kin terms.
- **I3** (learn): build paired evaluation sets differing only in the share of a target phenomenon, as a controlled probe alongside negative controls.
- **I4** (learn): report precision and recall errors per mention type for each repair rule or retraining, to catch improvements that actually add wrong links.
- **I5** (world): keep a persistent entity registry for the whole text, including name variants like "Rachel"/"Rach", so that later mentions attach to existing entities regardless of distance.

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

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,086 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,889 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works

🧭 Topics (1): [🗨️ Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · 401 works

## Cite

Cite as: Jaap Kruijt and Piek Vossen. 2022. The Role of Common Ground for Referential Expressions in Social Dialogues. Proceedings of the Fifth Workshop on Computational Models of Reference, Anaphora and Coreference.

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

```bibtex
@inproceedings{kruijt2022role,
  title = {{The Role of Common Ground for Referential Expressions in Social Dialogues}},
  author = {Jaap Kruijt and Piek Vossen},
  booktitle = {Proceedings of the Fifth Workshop on Computational Models of Reference, Anaphora and Coreference},
  year = {2022},
  doi = {10.48448/0nqw-fy37},
  url = {https://doi.org/10.48448/0nqw-fy37}
}
```

- [← previous The Influence Mechanism of High School English Grammar Science, Technology, Engineering, Art, and Mathematics Teaching Model on High School Students’ Learning Psychological Motivation](https://md.research.elamur.ai/free/en/papers/lin-2022-influence-mechanism-high-school-english/)
- [next → Comparing Native and Learner Englishes Using a Large Pre-trained Language Model](https://md.research.elamur.ai/free/en/papers/aoyama-2022-comparing-native-learner-englishes-large/)
- [↑ back to papers of 2022](https://md.research.elamur.ai/free/en/papers/2022/page/6/)

## References

20 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Jake Brawer, Olivier Mangin, Alessandro Roncone, Sarah Widder, and Brian Scassellati. 2018. Situated human–robot collaboration: predicting intent from grounded natural language. In 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 827–833. IEEE. [find](https://scholar.google.com/scholar?q=Jake+Brawer%2C+Olivier+Mangin%2C+Alessandro+Roncone%2C+Sarah+Widder%2C+and+Brian+Scassellati%2E+2018%2E+Situated+human%E2%80%93robot+collaboration%3A+predicting+intent+from+grounded+natural+language%2E+In+2018+IEEE%2FRSJ+Inter)
2. Joyce Y. Chai, Lanbo She, Rui Fang, Spencer Ottarson, Cody Littley, Changsong Liu, and Kenneth Hanson. 2014. Collaborative effort towards common ground in situated human-robot dialogue. In Proceedings of the 2014 ACM/IEEE International Conference on Human-Robot Interaction, HRI ’14, page 33–40, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Joyce+Y%2E+Chai%2C+Lanbo+She%2C+Rui+Fang%2C+Spencer+Ottarson%2C+Cody+Littley%2C+Changsong+Liu%2C+and+Kenneth+Hanson%2E+2014%2E+Collaborative+effort+towards+common+ground+in+situated+human%2Drobot+dialogue%2E+In+Proceedings)
3. Jinho D. Choi and Henry Y. Chen. 2018. SemEval 2018 task 4: Character identification on multiparty dialogues. In Proceedings of The 12th International Workshop on Semantic Evaluation, pages 57–64, New Orleans, Louisiana. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Jinho+D%2E+Choi+and+Henry+Y%2E+Chen%2E+2018%2E+SemEval+2018+task+4%3A+Character+identification+on+multiparty+dialogues%2E+In+Proceedings+of+The+12th+International+Workshop+on+Semantic+Evaluation%2C+pages+57%E2%80%9364%2C+New)
4. Sopan Khosla, Juntao Yu, Ramesh Manuvinakurike, Vincent Ng, Massimo Poesio, Michael Strube, and Carolyn Rosé. 2021. [find](https://scholar.google.com/scholar?q=Sopan+Khosla%2C+Juntao+Yu%2C+Ramesh+Manuvinakurike%2C+Vincent+Ng%2C+Massimo+Poesio%2C+Michael+Strube%2C+and+Carolyn+Ros%C3%A9%2E+2021%2E)
5. [The CODI-CRAC 2021 shared task on anaphora, bridging, and discourse deixis in dialogue](https://md.research.elamur.ai/free/en/papers/khosla-2021-codi-crac-2021-shared-task/). In Proceedings of the CODI-CRAC 2021 Shared Task on Anaphora, Bridging, and Discourse Deixis in Dialogue, pages 1–15, Punta Cana, Dominican Republic. Association for Computational Linguistics.
6. Sameer Pradhan, Alessandro Moschitti, Nianwen Xue, Olga Uryupina, and Yuchen Zhang. 2012. [Conll-2012 shared task: Modeling multilingual unrestricted coreference in ontonotes](https://md.research.elamur.ai/free/en/papers/pradhan-2012-conll-2012-shared-task-modeling/). In Joint Conference on EMNLP and CoNLL-Shared Task, pages 1–40.
7. Oliver Roesler and Ann Nowé. 2019. Action learning and grounding in simulated human–robot interactions. The Knowledge Engineering Review, 34. [↗](https://doi.org/10.1017/s0269888919000079)
8. Andy Shih, Arjun Sawhney, Jovana Kondic, Stefano Ermon, and Dorsa Sadigh. 2021. On the critical role of conventions in adaptive human-ai collaboration. arXiv preprint arXiv:2104.02871. [↗](https://arxiv.org/abs/2104.02871)
9. Herbert H Clark and Susan E Brennan. 1991. Grounding in communication. Perspectives on Socially Shared Cognition, pages 127–149. [↗](https://doi.org/10.1037/10096-006)
10. Darren Gergle, Robert E. Kraut, and Susan R. Fussell. 2013. Using visual information for grounding and awareness in collaborative tasks. Human–Computer Interaction, 28(1):1–39. [↗](https://doi.org/10.1080/07370024.2012.678246)
11. Janosch Haber, Tim Baumgärtner, Ece Takmaz, Lieke Gelderloos, Elia Bruni, and Raquel Fernández. 2019. [The photobook dataset: Building common ground through visually-grounded dialogue](https://md.research.elamur.ai/free/en/papers/haber-2019-photobook-dataset-building-common-ground/). arXiv preprint arXiv:1906.01530. [↗](https://arxiv.org/abs/1906.01530)
12. Robert D. Hawkins, Michael Franke, Michael C. Frank, Adele E. Goldberg, Kenny Smith, Thomas L. Griffiths, and Noah D. Goodman. 2021. From partners to populations: A hierarchical bayesian account of coordination and convention. [↗](https://doi.org/10.1037/rev0000348)
13. Mohit Shridhar and David Hsu. 2018. Interactive visual grounding of referring expressions for human-robot interaction. arXiv preprint arXiv:1806.03831. [↗](https://arxiv.org/abs/1806.03831)
14. Robert Stalnaker. 2002. Common ground. Linguistics and philosophy, 25(5/6):701–721. [find](https://scholar.google.com/scholar?q=Robert+Stalnaker%2E+2002%2E+Common+ground%2E+Linguistics+and+philosophy%2C+25%285%2F6%29%3A701%E2%80%93721%2E)
15. Xuewei Wang, Weiyan Shi, Richard Kim, Yoojung Oh, Sijia Yang, Jingwen Zhang, and Zhou Yu. 2019. [Persuasion for good: Towards a personalized persuasive dialogue system for social good](https://md.research.elamur.ai/free/en/papers/wang-2019-persuasion-good-personalized-persuasive-dialogue/).
16. Liyan Xu and Jinho D. Choi. 2020. [Revealing the myth of higher-order inference in coreference resolution](https://md.research.elamur.ai/free/en/papers/xu-2020-revealing-myth-higher-order-inference/). In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 8527–8533. Association for Computational Linguistics.
17. E. Holliman, J. Godfrey, and J. McDaniel. 1992. Switchboard: telephone speech corpus for research and development. In Acoustics, Speech, and Signal Processing, IEEE International Conference on, volume 1, pages 517–520, Los Alamitos, CA, USA. IEEE Computer Society. [find](https://scholar.google.com/scholar?q=E%2E+Holliman%2C+J%2E+Godfrey%2C+and+J%2E+McDaniel%2E+1992%2E+Switchboard%3A+telephone+speech+corpus+for+research+and+development%2E+In+Acoustics%2C+Speech%2C+and+Signal+Processing%2C+IEEE+International+Conference+on%2C+volume)
18. Mandar Joshi, Danqi Chen, Yinhan Liu, Daniel S. Weld, Luke Zettlemoyer, and Omer Levy. 2020. [Span-BERT: Improving Pre-training by Representing and Predicting Spans](https://md.research.elamur.ai/free/en/papers/joshi-2019-spanbert-improving-pre-training-representing/). Transactions of the Association for Computational Linguistics, 8:64–77.
19. Mandar Joshi, Omer Levy, Daniel S. Weld, and Luke Zettlemoyer. 2019. [BERT for coreference resolution: Baselines and analysis](https://md.research.elamur.ai/free/en/papers/joshi-2019-bert-coreference-resolution-baselines-analysis/). In Empirical Methods in Natural Language Processing (EMNLP).
20. Charles Kemp, Yang Xu, and Terry Regier. 2018. Semantic typology and efficient communication. Annual Review of Linguistics, 4(1):109–128. [↗](https://doi.org/10.1146/annurev-linguistics-011817-045406)

*Updated 24 October 2026*