Announcing "Proper Noun Proofreading AI" that Judges Terms in Context. A System Utilizing Internal Knowledge and Referencing a Company-Specific Dictionary Provided to Medical Book Publisher Nankodo.
Remedies Inc., operator of the AI proofreading service "wordrabbit," has launched a "Proper Noun Proofreading AI" for the medical publisher Nankodo. This advanced tool assists in professional-level quality control by having an AI interpret the contextual meaning of terms, rather than just performing simple character matching. It aims to prevent human oversight and enhance the quality and efficiency of medical book publishing, where absolute accuracy is paramount.
📋 Article Processing Timeline
- 📰 Published: April 15, 2026 at 22:15
- 🔍 Collected: April 15, 2026 at 14:01
- 🤖 AI Analyzed: April 15, 2026 at 20:07 (6h 6m after Collected)
Remedies Inc. (Headquarters: Chiyoda-ku, Tokyo; CEO: Go Fukuzawa), which operates the AI proofreading service "wordrabbit," has begun providing its "Proper Noun Proofreading AI" to Nankodo Co., Ltd. (Headquarters: Bunkyo-ku, Tokyo; hereinafter "Nankodo"), a publisher of specialized books and academic journals in a wide range of medical fields, including medicine, pharmacy, and nursing.
Instead of simple "character matching" like conventional proofreading tools, this secure system has been built to support advanced proofreading, with an AI that interprets the contextual meaning of terms. It is offered as a next-generation proofreading and editing infrastructure to back up professional-level quality control in the field of medical book publishing, where word-for-word accuracy is demanded.
**Challenges in Proofreading "Medical Terminology" in the Medical Field**
In the medical field, where word-for-word accuracy is directly linked to medical safety, errors in medical terms such as disease and drug names are unacceptable. However, the volume of medical terminology is vast, names vary widely by specialty, and it changes daily with the advancement of medicine. To date, the primary response has been visual confirmation by skilled editors and proofreaders, but there are limits to what can be checked by hand. Therefore, by utilizing AI, we aim to prevent human oversight, improve quality in medical book publishing, and advance production operations.
*A test manuscript is used, not an actual one.
**Comment from Nankodo**
Mr. Hiroaki Takeda, Director, Nankodo Co., Ltd.
Terminological accuracy is important in all publications, but it is especially crucial in the medical field. Giving shape to manuscripts entrusted to us by experts with accurate medical terminology is a key role of the editorial department. However, medical terminology is vast, and it is difficult for a single editor to grasp and proofread terms across all medical fields. For unfamiliar terms, it is necessary to check them one by one, consulting medical dictionaries, colleagues, the internet, and previously published books. This is a painstaking process that requires an enormous amount of time and effort. The "Proper Noun Proofreading AI," newly added to wordrabbit, is expected to reduce this burden and contribute significantly to quality improvement. By performing efficient screening based on our in-house medical terminology dictionary, our editors will be able to devote more of their time to "judgment." The expertise of our editors is the final bastion that guarantees the quality of our publications. By shortening work time through AI-powered proofreading, we can dedicate more time to the parts that require human judgment. As a result, we can release high-quality content to the world that makes the most of our editors' knowledge. The "Proper Noun Proofreading AI" will become an essential partner in our production process.
**How the wordrabbit "Proper Noun Proofreading AI" Works**
We have built a system where the AI references each company's unique "dictionary," which can contain hundreds of thousands of words, with every proofreading pass.
1. **An AI that thinks while consulting a dictionary**
The latest AI automatically and constantly references the "correct dictionary" for each client company. The AI does not make judgments based on its internal information alone but always cross-references the master data as its "evidence."
2. **Determines matches by "meaning," not just characters**
Conventional proper noun proofreading consists of simple text matching. It could not distinguish between terms that have similar characters but different meanings, such as "Ashikaga-shi (the clan)" and "Ashikaga-shi-hime (an individual)" in the text, and would flag them as errors.
With the new technology, the AI understands the meaning of terms and the surrounding context to distinguish and correctly judge such terms.
3. **Identifies proper and common nouns, eliminating unnecessary suggestions (noise)**
Conventional proper noun proofreading could not distinguish between proper and common nouns. As a result, it would compare words that simply sound similar, like "Rekovelle (proper noun)" and "level (common noun)," and flag it as an error.
The new technology accurately identifies and judges proper nouns.
*Rekovelle is a registered trademark of Ferring B.V.
4. **Automatically extracts proper nouns from the text**
It is possible to automatically extract proper noun candidates from the body of the text. This extraction function allows for the semi-automatic and continuous addition and updating of the dictionary in the future.
**Use Cases for This Function**
1. **Manufacturing manuals and specifications**
For quality control in environments where model numbers and part names are vast, and a single character error can lead to serious trouble or accidents.
2. **Medicine and healthcare**
For scenes where information accuracy is directly linked to social trust and safety, such as in pharmaceuticals, medical devices, and the standardization of medical data.
3. **Other domains with many proper nouns**
For fields with a large number of proper nouns and confirmation work, such as people's names or program titles.
**About "wordrabbit"**
"wordrabbit" is an AI solution specialized for proofreading Japanese text. In the proofreading process, which has traditionally been done entirely by hand, wordrabbit enhances professional quality control as a specialized AI that deeply interprets the nuances of the Japanese language and the rules of specialized domains. Unlike general-purpose chat AIs, it features a現場フィット型 (genba-fit) interface that can be seamlessly integrated into existing production flows, including proofreading of PDF data, which is essential in the publishing industry, and Microsoft Word.
https://wordrabbit.jp/
**About Nankodo Co., Ltd.**
Founded in 1879, Nankodo is a long-established publisher that has been publishing and selling specialized books and academic journals in a wide range of medical fields, including medicine, pharmacy, and nursing, for over 140 years. Its mission is to contribute to the advancement of medicine and healthcare by providing reliable information in the most suitable format for healthcare professionals and researchers. In recent years, it has also been actively developing digital content and online services.
**About Remedies Inc.**
Remedies is an AI startup company that creates solutions specialized in the Japanese language, based on its proprietary technology in "Japanese x AI." In 2024, it was selected as a finalist in one of Japan's largest startup pitch contests, "IVS LaunchPad," and in 2025, it was selected for the Tokyo Metropolitan Government's female venture growth promotion project, "APT Women." It has a partnership agreement with the National Institute for Japanese Language and Linguistics and also holds unique patents in the Japanese language domain.
Instead of simple "character matching" like conventional proofreading tools, this secure system has been built to support advanced proofreading, with an AI that interprets the contextual meaning of terms. It is offered as a next-generation proofreading and editing infrastructure to back up professional-level quality control in the field of medical book publishing, where word-for-word accuracy is demanded.
**Challenges in Proofreading "Medical Terminology" in the Medical Field**
In the medical field, where word-for-word accuracy is directly linked to medical safety, errors in medical terms such as disease and drug names are unacceptable. However, the volume of medical terminology is vast, names vary widely by specialty, and it changes daily with the advancement of medicine. To date, the primary response has been visual confirmation by skilled editors and proofreaders, but there are limits to what can be checked by hand. Therefore, by utilizing AI, we aim to prevent human oversight, improve quality in medical book publishing, and advance production operations.
*A test manuscript is used, not an actual one.
**Comment from Nankodo**
Mr. Hiroaki Takeda, Director, Nankodo Co., Ltd.
Terminological accuracy is important in all publications, but it is especially crucial in the medical field. Giving shape to manuscripts entrusted to us by experts with accurate medical terminology is a key role of the editorial department. However, medical terminology is vast, and it is difficult for a single editor to grasp and proofread terms across all medical fields. For unfamiliar terms, it is necessary to check them one by one, consulting medical dictionaries, colleagues, the internet, and previously published books. This is a painstaking process that requires an enormous amount of time and effort. The "Proper Noun Proofreading AI," newly added to wordrabbit, is expected to reduce this burden and contribute significantly to quality improvement. By performing efficient screening based on our in-house medical terminology dictionary, our editors will be able to devote more of their time to "judgment." The expertise of our editors is the final bastion that guarantees the quality of our publications. By shortening work time through AI-powered proofreading, we can dedicate more time to the parts that require human judgment. As a result, we can release high-quality content to the world that makes the most of our editors' knowledge. The "Proper Noun Proofreading AI" will become an essential partner in our production process.
**How the wordrabbit "Proper Noun Proofreading AI" Works**
We have built a system where the AI references each company's unique "dictionary," which can contain hundreds of thousands of words, with every proofreading pass.
1. **An AI that thinks while consulting a dictionary**
The latest AI automatically and constantly references the "correct dictionary" for each client company. The AI does not make judgments based on its internal information alone but always cross-references the master data as its "evidence."
2. **Determines matches by "meaning," not just characters**
Conventional proper noun proofreading consists of simple text matching. It could not distinguish between terms that have similar characters but different meanings, such as "Ashikaga-shi (the clan)" and "Ashikaga-shi-hime (an individual)" in the text, and would flag them as errors.
With the new technology, the AI understands the meaning of terms and the surrounding context to distinguish and correctly judge such terms.
3. **Identifies proper and common nouns, eliminating unnecessary suggestions (noise)**
Conventional proper noun proofreading could not distinguish between proper and common nouns. As a result, it would compare words that simply sound similar, like "Rekovelle (proper noun)" and "level (common noun)," and flag it as an error.
The new technology accurately identifies and judges proper nouns.
*Rekovelle is a registered trademark of Ferring B.V.
4. **Automatically extracts proper nouns from the text**
It is possible to automatically extract proper noun candidates from the body of the text. This extraction function allows for the semi-automatic and continuous addition and updating of the dictionary in the future.
**Use Cases for This Function**
1. **Manufacturing manuals and specifications**
For quality control in environments where model numbers and part names are vast, and a single character error can lead to serious trouble or accidents.
2. **Medicine and healthcare**
For scenes where information accuracy is directly linked to social trust and safety, such as in pharmaceuticals, medical devices, and the standardization of medical data.
3. **Other domains with many proper nouns**
For fields with a large number of proper nouns and confirmation work, such as people's names or program titles.
**About "wordrabbit"**
"wordrabbit" is an AI solution specialized for proofreading Japanese text. In the proofreading process, which has traditionally been done entirely by hand, wordrabbit enhances professional quality control as a specialized AI that deeply interprets the nuances of the Japanese language and the rules of specialized domains. Unlike general-purpose chat AIs, it features a現場フィット型 (genba-fit) interface that can be seamlessly integrated into existing production flows, including proofreading of PDF data, which is essential in the publishing industry, and Microsoft Word.
https://wordrabbit.jp/
**About Nankodo Co., Ltd.**
Founded in 1879, Nankodo is a long-established publisher that has been publishing and selling specialized books and academic journals in a wide range of medical fields, including medicine, pharmacy, and nursing, for over 140 years. Its mission is to contribute to the advancement of medicine and healthcare by providing reliable information in the most suitable format for healthcare professionals and researchers. In recent years, it has also been actively developing digital content and online services.
**About Remedies Inc.**
Remedies is an AI startup company that creates solutions specialized in the Japanese language, based on its proprietary technology in "Japanese x AI." In 2024, it was selected as a finalist in one of Japan's largest startup pitch contests, "IVS LaunchPad," and in 2025, it was selected for the Tokyo Metropolitan Government's female venture growth promotion project, "APT Women." It has a partnership agreement with the National Institute for Japanese Language and Linguistics and also holds unique patents in the Japanese language domain.