JSA

What JSA does

JSA (株式会社日本システムアプリケーション / Japan System Applications Co., Ltd.) is a privately held, 22-employee Japanese engineering company headquartered in Toshima-ku, Tokyo, founded in September 1969 with registered capital of 24 million JPY. The company operates three integrated businesses: (1) custom system and hardware development for research institutions and enterprises, (2) a Japanese natural language processing (NLP) product line built on the EDR (Electronic Dictionary and Resources) Japanese dictionary framework originally developed by NICT (National Institute of Information and Communications Technology), and (3) language data product distribution alongside ML/LLM training-data annotation services.

Its core NLP products are EdrMAMe (a morphological analyzer using the EDR Japanese word dictionary with ~12,000 proprietary added entries and concept identifier linking), Ko-BaKo/J (a multi-functional Japanese language processing Windows library with ~1.5M vocabulary entries and semantic attribute tagging including AKO, THISA, THPOF, THVAL), Ko-BaKo/E (an equivalent English-language processing library), and 文-AYA- (a semantic analysis tool performing morphological, dependency, and deep semantic parsing using ~400K EDR entries with 30 deep-case labels and tense/aspect/modality handling, derived from 15+ years of university research). The company also offers TransTalking (a Japanese-English translation Android app), specialized hardware (real-time vehicle trajectory recording systems; VSDR800 VTR devices — now discontinued), and machine learning annotation and fine-tuning services using BERT/Transformer, PyTorch, and Scikit-learn.

JSA's business model mixes one-time/perpetual software licensing (Ko-BaKo/J at 1.1M JPY per department or 550K JPY academic with 5% GSK member discount), channel-distributed language data products (Sanshindo Publishers for the Japanese verb complementarity dictionary; NICT-authorized reseller for EDR dictionaries), and bespoke professional services for enterprise and research clients. Its go-to-market is sales-led and direct — products are sold through the jsa.co.jp website with product-specific contact forms — and the company serves a diversified customer base spanning national research institutes (NICT, RIKEN, NII, NINJAL), leading Japanese universities (Tokyo, Kyoto, Tokyo Tech, Tsukuba, Tohoku, Otemon Gakuin), and major enterprises including multiple NEC group companies, Honda R&D, Mitsubishi Motors, Fujifilm, and LIXIL. The firm maintains a technology blog (tech.jsa.co.jp) for thought leadership and runs an active recruitment page, signaling continued investment despite its small absolute scale.

Name

JSA

Legal name

株式会社日本システムアプリケーション

Website

https://jsa.co.jp

Company type

Private

Founded year

1969

Operating status

Operating

Headcount range

11–50 employees

Short description

JSA (Japan System Applications Co., Ltd.) is a 22-employee, privately held Japanese engineering firm founded in 1969 that develops Japanese natural language processing software (Ko-BaKo/J, EdrMAMe, AYA) built on NICT's EDR dictionaries, along with custom software and hardware systems for universities, national research institutes, and Japanese enterprises.

Ownership category

akta.pro rank

Product category

Japanese Natural Language Processing Software

NAICS

Software Publishers (5132), Custom Computer Programming Services (541511)

SIC

Services-Prepackaged Software (7372), Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

NLP Developer Tools & Evaluation (prompting, testing, benchmarks) (HDAAADAK)

akta.pro secondary industry

Machine Translation & Multilingual NLP (HDAAADAD)

Keywords

• Japanese natural language processing
• Morphological analysis software
• Semantic parsing tools
• Language data dictionaries
• Custom system development

Headquarters

HQ city

Toshima

HQ country

Japan

HQ region

Asia

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure

Revenue model

1. Software Product Licensing (Ko-BaKo/J, Ko-BaKo/E): Sale of Japanese and English language processing DLL libraries. Ko-BaKo/J is sold as a perpetual license for 1.1 million JPY (tax included) per unit/department (up to 10 seats). Academic pricing is 550,000 JPY (tax included) per research lab (up to 10 seats). GSK members receive a 5% discount.
2. Language Data Products (日本語動詞の結合価, 未知語集, EDR辞書): Sale of specialized language data products. The Japanese verb complementarity dictionary (結合価データ) is published and sold through Sanshindo publishers. EDR dictionaries are sold on behalf of NICT. Unknown word lists and concept system editing tools are sold as add-on/supplementary products.
3. Custom System Development (システム開発・研究開発支援): Custom software and hardware development services for research institutions and corporations. Includes Web systems, smartphone applications, Windows/Linux applications, text mining tools, translation tools, and hardware development. Also includes research and development support services using cutting-edge technologies (voice, video, location-based, network).
4. Machine Learning Annotation Services: Annotation of training data for machine learning and large language models, development of ML-based systems, and model fine-tuning services. Part of the data creation and analysis business unit.
5. Hardware Product Sales (VSDR800, リアルタイム走行車両軌跡記録装置): Sale of specialized hardware including VSDR800 VTR data collection/playback devices and real-time vehicle trajectory recording systems. Note: VSDR800 is listed as discontinued/sales ended.

Pricing tiers

Go-to-market motion1 record

Distribution channels3 records

Marketing channels3 records

Core offering

JSA develops and sells proprietary Japanese natural language processing software (Ko-BaKo/J, Ko-BaKo/E, EdrMAMe, AYA) and distributes specialized language data products (EDR dictionaries, verb complementarity dictionary), while providing custom software/hardware system development, machine learning annotation services, and the TransTalking translation app for research institutions, universities, and Japanese enterprises.

Differentiator

Problem solved

Functional benefit

Products and services

• Ko-BaKo/J (Multi-functional Japanese Language Processing Library) Multi-functional Japanese language processing library for Windows providing morphological analysis with part-of-speech and semantic attribute tagging (AKO, THISA, THPOF, THVAL), syntactic parsing with dependency structure and case-role analysis, and user dictionary customization. Built on ~1.5 million vocabulary entries. Sold as a perpetual license for 1,100,000 JPY per department (up to 10 seats), with academic pricing of 550,000 JPY per research lab.
• Ko-BaKo/E (Multi-functional English Language Processing Library) Multi-functional English language processing library for Windows, providing linguistic analysis capabilities similar to Ko-BaKo/J for English text. Used by research institutions and corporate customers requiring English NLP alongside the company's Japanese offerings.
• EdrMAMe (EDR Dictionary-based Morphological Analyzer) Morphological analyzer using the EDR Japanese word dictionary that outputs morphological information plus concept identifiers linking to the EDR concept dictionary. Handles unknown words, assigns part-of-speech tags, left/right connection attributes, and concept IDs. Adds 12,000 words and excludes ~400 inappropriate single-word entries from the EDR dictionary. Dictionary size 350MB; compatible with Windows 10 and 11.
• AYA (文-AYA- Japanese Semantic Analysis Tool) Japanese semantic analysis tool performing morphological segmentation, dependency parsing, and semantic parsing. Leverages 15+ years of university research. Outputs semantic graph structures with concept identification from ~400,000 EDR dictionary entries and deep case roles from 30 EDR/independently defined labels. Handles tense, aspect, and modality analysis (e.g., distinguishing 'taberu', 'tabeta', 'tabetai').
• TransTalking (Japanese-English Translation App with Speech Synthesis) Japanese-English translation application providing speech recognition for voice input and speech synthesis (text-to-speech) for read-aloud functionality. Supports offline use by downloading language data. Available as an Android app and processes images for text recognition.
• EDR Electronic Dictionaries (EDR Japanese Word Dictionary and Concept Dictionary) EDR (Electronic Dictionary and Resources) Japanese word dictionaries and concept dictionaries originally developed by NICT (National Institute of Information and Communications Technology). JSA handles sales and distribution of these dictionaries on behalf of NICT within Japan, serving research institutions, universities, and enterprise customers requiring comprehensive Japanese linguistic resources.
• Japanese Verb Complementarity Dictionary (日本語動詞の結合価) Specialized Japanese verb complementarity (valency) dictionary derived from real examples in the EDR corpus. Published as a commercial book with CD-ROM containing search software, distributed through Sanshindo publishers (株式会社三省堂). Sold via book retailers at publisher-determined pricing.
• Custom System Development Services (システム開発) Custom software and hardware development services for research institutions and corporations. Includes Web systems, smartphone applications, Windows/Linux applications, text mining tools, translation tools, and specialized hardware development. Also includes research and development support services using cutting-edge technologies (voice, video, location-based, network).
• Machine Learning Annotation Services (機械学習アノテーション業務) Annotation of training data for machine learning and large language models, development of ML-based systems using BERT/Transformer, PyTorch, and Scikit-learn, and model fine-tuning services for accuracy improvement. Supports applications in semantic disambiguation, concept ID refinement, and NLP model training.
• Real-time Vehicle Trajectory Recording Device (リアルタイム走行車両軌跡記録装置) Specialized hardware device that records real-time vehicle trajectory data for automotive and industrial data collection applications. Used by automotive manufacturers and research customers for vehicle dynamics analysis.

Quantifiable outcome

• EdrMAMe dictionary size: 350MB with 12,000 added vocabulary entries and ~400 excluded single-word entries
• +2 more outcomes

Named customers15 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability9 records

Feature5 records

Partnerships

Five partnerships are on record, tiered core and minor.

• 株式会社三省堂 (Sanshindo Publishers)coreChannel Partner/ Reseller/ Distributor · 1 November 2003JSA developed the Japanese verb complementarity dictionary (結合価データ) from EDR corpus real examples, which Sanshindo publishes and distributes as a commercial book with CD-ROM. This represents a long-term distribution partnership where JSA provides content expertise and Sanshindo handles publication and retail distribution.
• 国立研究開発法人情報通信研究機構 (NICT) / 独立行政法人情報処理推進機構 (IPA)coreStrategic or Co-development PartnerJSA uses and distributes EDR (Electronic Dictionary and Resources) Japanese word dictionaries and concept dictionaries developed by NICT. JSA also participates in IPA's SECURITY ACTION program (two-star certification), demonstrating alignment with national information security initiatives. EDR dictionaries form the foundation of JSA's core NLP technology products.
• 言語資源協会 (GSK - Language Resource Association)minorChannel Partner/ Reseller/ DistributorGSK members receive a 5% discount on Ko-BaKo/J products. GSK is a language resource industry association whose members are eligible for preferential pricing on JSA's language processing tools.
• Honda R&D / Honda Research Institute JapanminorOthersListed as a major client/customer. JSA provides custom system development, NLP services, and potentially hardware solutions to Honda's research divisions.
• NEC Group CompaniesminorOthersMultiple NEC group companies (NEC, NEC Solution Innovator, NEC Networks & Sensors, NEC BIZINT, NEC Fielding, NEC Platforms) are listed as major clients/customers for JSA's system development and language processing services.

Scale indicators3 records

Recent moves5 records

Expansion highlights4 records

Direct peers

• NEC Corporation: NEC develops its own Japanese NLP platform and is also one of JSA's largest enterprise customers. NEC's in-house NLP competes with JSA's products for Japanese-language AI buyers while sharing enterprise customers.
• Appen Limited: Appen provides ML training data annotation and data services that directly overlap with JSA's machine learning annotation business, with both targeting NLP/LLM training data pipelines at scale.
• NTT (NTT Human Informatics Laboratories): NTT's research arm builds Japanese NLP tools, semantic analysis, and machine translation technologies that compete directly with JSA's Ko-BaKo/J and AYA products for the same academic and enterprise research buyers in Japan.
• Recruit (Workwise): Recruit operates Workwise and other Japanese NLP-powered HR/recruitment products, representing a direct competitor in commercial Japanese NLP applications serving Japanese enterprises with semantic analysis capabilities.

Broad incumbents

• Fujitsu Limited: Fujitsu operates a major Japanese NLP/AI program (including Zinrai) that overlaps with JSA's enterprise services across the same Japanese corporate customer base, though as part of a much broader IT services portfolio.
• Toshiba Corporation: Toshiba's digital solutions division develops Japanese NLP and text mining technologies, placing it in the same broad competitive set as JSA for Japanese enterprise and research NLP buyers.
• Hitachi Ltd. Hitachi offers Japanese-language AI and text analytics capabilities within its broader Lumada platform, overlapping with JSA's enterprise NLP system development for large Japanese manufacturing and services customers.

Others

• NICT (National Institute of Information and Communications Technology): NICT is the upstream developer of the EDR dictionaries that JSA commercializes and builds upon. It is both JSA's most important strategic partner and the supplier of the core data assets JSA's products depend on — making it a critical ecosystem participant rather than a competitor.

Emerging players

• DeepL: DeepL offers Japanese-English machine translation that competes with JSA's TransTalking product for translation use cases, and is a benchmark for next-generation translation quality that JSA must differentiate against.
• Scale AI: Scale AI provides ML annotation and fine-tuning services for foundation models, intersecting with JSA's LLM/NLP annotation offering — especially for high-quality Japanese-language training data where Scale is building but has historically been weaker.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights7 records

Customer concentration

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about JSA