---
title: "JPPB: Automatic Construction of a — Nishiyama 2026"
description: "People describing their own symptoms seldom use clinical vocabulary."
authors: ["Tomohiro Nishiyama", "Mana Kuramoto", "Shoko Wakamiya", "Eiji ARAMAKI"]
year: 2026
venue: "Proceedings of the Fifteenth Language Resources and Evaluation Conference"
source: "https://aclanthology.org/2026.lrec-1.621/"
doi: "10.63317/345uq5t7y98h"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/nishiyama-2026-jppb-automatic-construction-soft-labeled/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/nishiyama-2026-jppb-automatic-construction-soft-labeled/)

[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/6/) / JPPB: Automatic Construction of a Soft-Labeled Japanese Patient Phrase Bank for Symptom Normalization

# JPPB: Automatic Construction of a Soft-Labeled Japanese Patient Phrase Bank for Symptom Normalization

**Authors:** [Tomohiro Nishiyama](https://md.research.elamur.ai/free/en/authors/tomohiro-nishiyama/), Mana Kuramoto, [Shoko Wakamiya](https://md.research.elamur.ai/free/en/authors/shoko-wakamiya/), [Eiji ARAMAKI](https://md.research.elamur.ai/free/en/authors/eiji-aramaki/) · **Year:** 2026 · **Venue:** Proceedings of the Fifteenth Language Resources and Evaluation Conference · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.lrec-1.621/) · [PDF](https://aclanthology.org/2026.lrec-1.621.pdf) · [DOI 10.63317/345uq5t7y98h](https://doi.org/10.63317/345uq5t7y98h) · [OpenAlex](https://openalex.org/W7160076052)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14)

## Summary

People describing their own symptoms seldom use clinical vocabulary. Japanese patients in particular rely on onomatopoeia and context-dependent adjectives, so mapping their words to standard symptom terms is hard. Existing Japanese patient lexicons are curated by hand, work at the word level, and assign one term per entry. The authors built the Japanese Patient Phrase Bank (JPPB) without manual curation. They took symptom terms from JMED-DICT and had GPT-4.1-mini, prompted with different patient personas, write colloquial first-person phrases for each term. These phrases were then given soft labels: each one receives a weighted set of candidate canonical terms, based on which terms its nearest embedding neighbours point to. The final resource contains 7,035 phrase–term pairs covering 412 symptoms. On the KEEPHA and MedNLP-SC datasets, soft labels gave better Top-1 accuracy than hard labels and matched gold label distributions more closely. LLM-based normalization still performed best overall. Even so, JPPB is a transparent, lightweight option that can run locally, which matters where privacy rules limit the use of external APIs.

## Key points

- **Construction pipeline:**
  - Source terms are 783 surface/canonical pairs from JMED-DICT, limited to ICD-10 "R" codes (symptoms and signs).
  - GPT-4.1-mini wrote three phrases per term for each of three personas: general adult, child, and elderly.
- **Soft labeling:**
  - For each phrase, the 10 most similar phrases are retrieved with a multilingual MiniLM model.
  - Every canonical term linked to those neighbours adds 0.1 to that term's score, so the maximum score is 1.0.
- **Resource statistics:**
  - In the soft-label version, a canonical term links to more than 96 variants on average, with a maximum of 1,015.
  - In the hard-label version, the average is 17 variants and the maximum is 261.
  - Label entropy has a minimum of 0 and a maximum of 2.303; its average is 1.483 and its median is 1.557. Most phrases therefore map to roughly 2–5 plausible terms.
- **Evaluation:** On KEEPHA, Top-1 recall was 0.250 with soft labels, against 0.075 with hard labels and 0.188 for the manual baseline dictionary. The baseline dictionary usually had the highest precision.
- **Availability:** JPPB is released under CC-BY at github.com/Tomohiro-git/JPPB.

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

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cite

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

```bibtex
@inproceedings{nishiyama2026jppb,
  title = {{JPPB: Automatic Construction of a Soft-Labeled Japanese Patient Phrase Bank for Symptom Normalization}},
  author = {Nishiyama, Tomohiro and Kuramoto, Mana and Wakamiya, Shoko and ARAMAKI, Eiji},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/345uq5t7y98h},
  url = {https://doi.org/10.63317/345uq5t7y98h}
}
```

- [← previous JobArabi: An Arabic Corpus and Analysis of Job Announcements from Social Media](https://md.research.elamur.ai/free/en/papers/zaghouani-2026-jobarabi-arabic-corpus-analysis-job/)
- [next → JSTS-Neg: Japanese Semantic Textual Similarity Dataset for Evaluating Negation Understanding Ability](https://md.research.elamur.ai/free/en/papers/yuasa-2026-jsts-neg-japanese-semantic-textual/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/6/)

## References

20 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Fabio Dennstädt, Janna Hastings, Paul Martin Putora, Max Schmerder, and Nikola Cihoric. 2025. Implementing large language models in healthcare while balancing control, collaboration, costs and security. npj Digital Medicine, 8(1):143. [find](https://scholar.google.com/scholar?q=Fabio+Dennst%C3%A4dt%2C+Janna+Hastings%2C+Paul+Martin+Putora%2C+Max+Schmerder%2C+and+Nikola+Cihoric%2E+2025%2E+Implementing+large+language+models+in+healthcare+while+balancing+control%2C+collaboration%2C+costs+and+securit)
2. Katrin Erk, Diana McCarthy, and Nicholas Gaylord. 2009. Investigations on word senses and word usages. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP, pages 10–18, Suntec, Singapore.… [find](https://scholar.google.com/scholar?q=Katrin+Erk%2C+Diana+McCarthy%2C+and+Nicholas+Gaylord%2E+2009%2E+Investigations+on+word+senses+and+word+usages%2E+In+Proceedings+of+the+Joint+Conference+of+the+47th+Annual+Meeting+of+the+ACL+and+the+4th+Internat)
3. Gen Gu, Xingting Zhang, Xingeng Zhu, Zhe Jian, Ken Chen, Dong Wen, Li Gao, Shaodian Zhang, Fei Wang, Handong Ma, and Jianbo Lei. 2019. Development of a Consumer Health Vocabulary by Mining Health Forum Texts Based on Word Embedding: Semiautomatic Approach. JMIR Medical Informatics, 7(2):e12704. [find](https://scholar.google.com/scholar?q=Gen+Gu%2C+Xingting+Zhang%2C+Xingeng+Zhu%2C+Zhe+Jian%2C+Ken+Chen%2C+Dong+Wen%2C+Li+Gao%2C+Shaodian+Zhang%2C+Fei+Wang%2C+Handong+Ma%2C+and+Jianbo+Lei%2E+2019%2E+Development+of+a+Consumer+Health+Vocabulary+by+Mining+Health+Foru)
4. E. Hayes, R. Dua, E. Yeung, and K. Fan. 2017. Patient understanding of commonly used oral medicine terminology. British Dental Journal, 223(11):842–845. [find](https://scholar.google.com/scholar?q=E%2E+Hayes%2C+R%2E+Dua%2C+E%2E+Yeung%2C+and+K%2E+Fan%2E+2017%2E+Patient+understanding+of+commonly+used+oral+medicine+terminology%2E+British+Dental+Journal%2C+223%2811%29%3A842%E2%80%93845%2E)
5. Mohammed Ibrahim, Susan Gauch, Omar Salman, and Mohammed Alqahtani. 2021. An automated method to enrich consumer health vocabularies using GloVe word embeddings PeerJ. and an auxiliary lexical resource. [find](https://scholar.google.com/scholar?q=Mohammed+Ibrahim%2C+Susan+Gauch%2C+Omar+Salman%2C+and+Mohammed+Alqahtani%2E+2021%2E+An+automated+method+to+enrich+consumer+health+vocabularies+using+GloVe+word+embeddings+PeerJ%2E+and+an+auxiliary+lexical+resourc)
6. Kaoru Ito, Hiroyuki Nagai, Taro Okahisa, Shoko Wakamiya, Tomohide Iwao, and Eiji Aramaki. 2018. J-MeDic: A Japanese disease name dictionary based on real clinical usage. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. Europea… [find](https://scholar.google.com/scholar?q=Kaoru+Ito%2C+Hiroyuki+Nagai%2C+Taro+Okahisa%2C+Shoko+Wakamiya%2C+Tomohide+Iwao%2C+and+Eiji+Aramaki%2E+2018%2E+J%2DMeDic%3A+A+Japanese+disease+name+dictionary+based+on+real+clinical+usage%2E+In+Proceedings+of+the+Eleventh)
7. Sarvnaz Karimi, Alejandro Metke-Jimenez, Madonna Kemp, and Chen Wang. 2015. Cadec: A corpus of adverse drug event annotations. Journal of Biomedical Informatics, 55:73–81. [find](https://scholar.google.com/scholar?q=Sarvnaz+Karimi%2C+Alejandro+Metke%2DJimenez%2C+Madonna+Kemp%2C+and+Chen+Wang%2E+2015%2E+Cadec%3A+A+corpus+of+adverse+drug+event+annotations%2E+Journal+of+Biomedical+Informatics%2C+55%3A73%E2%80%9381%2E)
8. Alla Keselman, Catherine Arnott Smith, Guy Divita, Hyeoneui Kim, Allen C. Browne, Gondy Leroy, and Qing Zeng-Treitler. 2008. Consumer health concepts that do not map to the UMLS: where do they fit? Journal of the American Medical Informatics Association: JAMIA, 15(4):496–505. [find](https://scholar.google.com/scholar?q=Alla+Keselman%2C+Catherine+Arnott+Smith%2C+Guy+Divita%2C+Hyeoneui+Kim%2C+Allen+C%2E+Browne%2C+Gondy+Leroy%2C+and+Qing+Zeng%2DTreitler%2E+2008%2E+Consumer+health+concepts+that+do+not+map+to+the+UMLS%3A+where+do+they+fit%3F+Jo)
9. Mihiro Nishidani, Shuntaro Yada, Shoko Wakamiya, and Eiji Aramaki. 2021. 生成 アプローチによる患者表現の標準化 [standardization of patient expressions via a generative approach] (in japanese). In Proceedings of the JSAI Special Interest Group on Artificial Intelligence in Medicine (SIG-AIMED). The Japanese Society fo… [find](https://scholar.google.com/scholar?q=Mihiro+Nishidani%2C+Shuntaro+Yada%2C+Shoko+Wakamiya%2C+and+Eiji+Aramaki%2E+2021%2E+%E7%94%9F%E6%88%90+%E3%82%A2%E3%83%97%E3%83%AD%E3%83%BC%E3%83%81%E3%81%AB%E3%82%88%E3%82%8B%E6%82%A3%E8%80%85%E8%A1%A8%E7%8F%BE%E3%81%AE%E6%A8%99%E6%BA%96%E5%8C%96+%5Bstandardization+of+patient+expressions+via+a+generative+approach%5D+%28in+japanese%29%2E+In+Proceedings+of+the+JSA)
10. Tomohiro Nishiyama, Ayane Yamaguchi, Peitao Han, Lis Weiji Kanashiro Pereira, Yuka Otsuki, Gabriel Herman Bernardim Andrade, Noriko Kudo, Shuntaro Yada, Shoko Wakamiya, Eiji Aramaki, Masahiro Takada, and Masakazu Toi. 2024. Automated System to Capture Patient Symptoms From Multitype Japanese Clinica… [find](https://scholar.google.com/scholar?q=Tomohiro+Nishiyama%2C+Ayane+Yamaguchi%2C+Peitao+Han%2C+Lis+Weiji+Kanashiro+Pereira%2C+Yuka+Otsuki%2C+Gabriel+Herman+Bernardim+Andrade%2C+Noriko+Kudo%2C+Shuntaro+Yada%2C+Shoko+Wakamiya%2C+Eiji+Aramaki%2C+Masahiro+Takada%2C+)
11. Yukiko Ohno, Riri Kato, Haruki Ishikawa, Tomohiro Nishiyama, Minae Isawa, Mayumi Mochizuki, Eiji Aramaki, and Tohru Aomori. 2024. Using the Natural Language Processing System Medical Named Entity Recognition-Japanese to Analyze Pharmaceutical Care Records: Natural Language Processing Analysis. JMIR… [find](https://scholar.google.com/scholar?q=Yukiko+Ohno%2C+Riri+Kato%2C+Haruki+Ishikawa%2C+Tomohiro+Nishiyama%2C+Minae+Isawa%2C+Mayumi+Mochizuki%2C+Eiji+Aramaki%2C+and+Tohru+Aomori%2E+2024%2E+Using+the+Natural+Language+Processing+System+Medical+Named+Entity+Reco)
12. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Nils+Reimers+and+Iryna+Gurevych%2E+2019%2E+Sentence%2Dbert%3A+Sentence+embeddings+using+siamese+bert%2Dnetworks%2E+In+Proceedings+of+the+2019+Conference+on+Empirical+Methods+in+Natural+Language+Processing%2E+Associ)
13. Abeed Sarker, Rui Zhang, Yanshan Wang, Yunyu Xiao, Sudeshna Das, Dalton Schutte, David Oniani, Qianqian Xie, and Hua Xu. 2024. Natural Language Processing for Digital Health in the Era of Large Language Models. Yearbook of Medical Informatics, 33(01):229–240. [find](https://scholar.google.com/scholar?q=Abeed+Sarker%2C+Rui+Zhang%2C+Yanshan+Wang%2C+Yunyu+Xiao%2C+Sudeshna+Das%2C+Dalton+Schutte%2C+David+Oniani%2C+Qianqian+Xie%2C+and+Hua+Xu%2E+2024%2E+Natural+Language+Processing+for+Digital+Health+in+the+Era+of+Large+Langua)
14. Zonghai Yao, Nandyala Siddharth Kantu, Guanghao Wei, Hieu Tran, Zhangqi Duan, Sunjae Kwon, Zhichao Yang, and Hong Yu. 2024. README: Bridging Medical Jargon and Lay Understanding for Patient Education through Data-Centric NLP. In Findings of the Association for Computational Linguistics: EMNLP 2024,… [find](https://scholar.google.com/scholar?q=Zonghai+Yao%2C+Nandyala+Siddharth+Kantu%2C+Guanghao+Wei%2C+Hieu+Tran%2C+Zhangqi+Duan%2C+Sunjae+Kwon%2C+Zhichao+Yang%2C+and+Hong+Yu%2E+2024%2E+README%3A+Bridging+Medical+Jargon+and+Lay+Understanding+for+Patient+Education+)
15. Q. Zeng, S. Kogan, N. Ash, R. A. Greenes, and A. A. Boxwala. 2002. Characteristics of Consumer Terminology for Health Information Retrieval. Methods of Information in Medicine, 41(04):289–298. [find](https://scholar.google.com/scholar?q=Q%2E+Zeng%2C+S%2E+Kogan%2C+N%2E+Ash%2C+R%2E+A%2E+Greenes%2C+and+A%2E+A%2E+Boxwala%2E+2002%2E+Characteristics+of+Consumer+Terminology+for+Health+Information+Retrieval%2E+Methods+of+Information+in+Medicine%2C+41%2804%29%3A289%E2%80%93298%2E)
16. Qing T. Zeng and Tony Tse. 2006. Exploring and developing consumer health vocabularies. Journal of the American Medical Informatics Association: JAMIA, 13(1):24–29. [find](https://scholar.google.com/scholar?q=Qing+T%2E+Zeng+and+Tony+Tse%2E+2006%2E+Exploring+and+developing+consumer+health+vocabularies%2E+Journal+of+the+American+Medical+Informatics+Association%3A+JAMIA%2C+13%281%29%3A24%E2%80%9329%2E)
17. Zeng Qing, Kogan Sandra, Ash Nachman, and Greenes Robert A. 2001. Patient and Clinician Vocabulary: How Different Are They? In Studies in Health Technology and Informatics. IOS Press. [find](https://scholar.google.com/scholar?q=Zeng+Qing%2C+Kogan+Sandra%2C+Ash+Nachman%2C+and+Greenes+Robert+A%2E+2001%2E+Patient+and+Clinician+Vocabulary%3A+How+Different+Are+They%3F+In+Studies+in+Health+Technology+and+Informatics%2E+IOS+Press%2E)
18. Mingqian Zheng, Jiaxin Pei, Lajanugen Logeswaran, Moontae Lee, and David Jurgens. 2024. When ” A Helpful Assistant” Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages… [find](https://scholar.google.com/scholar?q=Mingqian+Zheng%2C+Jiaxin+Pei%2C+Lajanugen+Logeswaran%2C+Moontae+Lee%2C+and+David+Jurgens%2E+2024%2E+When+%E2%80%9D+A+Helpful+Assistant%E2%80%9D+Is+Not+Really+Helpful%3A+Personas+in+System+Prompts+Do+Not+Improve+Performances+of+Lar)
19. Maryam Zolnoori, Kin Wah Fung, Timothy B. Patrick, Paul Fontelo, Hadi Kharrazi, Anthony Faiola, Nilay D. Shah, Yi Shuan Shirley Wu, Christina E. Eldredge, Jake Luo, Mike Conway, Jiaxi Zhu, Soo Kyung Park, Kelly Xu, and Hamideh Moayyed. 2019. The PsyTAR dataset: From patients generated narratives to… [find](https://scholar.google.com/scholar?q=Maryam+Zolnoori%2C+Kin+Wah+Fung%2C+Timothy+B%2E+Patrick%2C+Paul+Fontelo%2C+Hadi+Kharrazi%2C+Anthony+Faiola%2C+Nilay+D%2E+Shah%2C+Yi+Shuan+Shirley+Wu%2C+Christina+E%2E+Eldredge%2C+Jake+Luo%2C+Mike+Conway%2C+Jiaxi+Zhu%2C+Soo+Kyung+P)
20. Shoko Wakamiya, Lis Kanashiro Pereira, Lisa Raithel, Katherine Yeh, Peitao Han, Seiji Shimizu, Tomohiro Nishiyama, Gabriel Herman Bernardim Andrade, Noriki Nishida, Hiroki Teranishi, Narumi Tokunaga, Philippe Thomas, Roland Roller, Pierre Zweigenbaum, Yuji Matsumoto, Akiko Aizawa, Sebastian Möller,… [find](https://scholar.google.com/scholar?q=Shoko+Wakamiya%2C+Lis+Kanashiro+Pereira%2C+Lisa+Raithel%2C+Katherine+Yeh%2C+Peitao+Han%2C+Seiji+Shimizu%2C+Tomohiro+Nishiyama%2C+Gabriel+Herman+Bernardim+Andrade%2C+Noriki+Nishida%2C+Hiroki+Teranishi%2C+Narumi+Tokunaga%2C+)