Japan Industrial PhD Matching Support System Development

Background & Challenge

The University of Electro-Communications (UEC) is building a matching support platform to advance industry-academia doctoral programs (Japan's version of the "Industrial PhD"), accurately connecting the research needs of companies and municipalities with the expertise and research achievements of university faculty and graduate students. Because research needs and achievements are described in free-form natural language, "vocabulary mismatch" — synonyms, abbreviations, paraphrasing, and spelling variation — made it difficult for keyword search alone to achieve sufficient matching accuracy. This system has been developed as a platform used by four types of participants: companies/municipalities, faculty members, graduate students, and coordinators.

Udzuki's Role

Designed a matching approach that captures how closely research themes relate in meaning

Compared several candidate approaches and adopted the one best suited to capturing similarity between research needs and achievements — including paraphrasing and wording variation that simple keyword matching cannot catch

Implemented semantic search that stays fast even at scale

Implemented a meaning-based search mechanism that keeps response times steady as the number of research projects, faculty, and students grows

Built a web platform supporting four user types

Developed onboarding, profile management, and home screens for companies/municipalities, faculty members, graduate students, and coordinators, end to end from requirements onward

Implemented automatic processing triggered by data updates

Implemented a mechanism that detects new or updated research projects, profiles, and publications and automatically refreshes the information used for matching

Implemented three types of matching

Implemented matching between projects and faculty, projects and students, and students and faculty, together with a configurable similarity threshold and a way to review execution history