In a remarkable achievement for Japanese higher education, Rikiya Takehi, a recent undergraduate from Waseda University's School of Fundamental Science and Engineering, has secured the Best Full Research Paper Award at the prestigious ACM WSDM 2026 conference. This top honor, awarded to just one paper out of 799 submissions, underscores the growing prowess of Japan's young AI researchers and highlights Waseda University's commitment to fostering elite talent in information retrieval and machine learning.
The Web Search and Data Mining (WSDM) conference, recognized as an A*-ranked event by the Computing Research and Education Association of Australasia (CORE), draws top global experts to advance technologies powering modern search engines. Held in the United States in February 2026, it featured rigorous peer review with an acceptance rate of approximately 16%. Takehi's solo-led paper, "Diversification as Risk Minimization," stood out for its innovative mathematical framework addressing longstanding challenges in search result diversification.
The Innovation Behind the Award-Winning Paper
Search engines like Google or Bing frequently encounter ambiguous queries—think "Apple," which could refer to the fruit or the tech giant. Traditional diversification algorithms aim to balance results across multiple user intents, but they often fall short, favoring popular interpretations and neglecting minority intents. Takehi's research reframes diversification as a risk minimization problem, introducing VRisk, a metric that quantifies the expected dissatisfaction for the least-served intents.
By optimizing VRisk through simple mathematical transformations and empirical validation on benchmarks like TREC Web tracks, the approach produces more robust rankings. Experiments demonstrated superior performance over state-of-the-art methods, proving diversification's inherent limitations akin to relevance-only ranking. Collaborators including Fernando Diaz from Carnegie Mellon University and experts from Cornell University and the U.S. National Institute of Standards and Technology (NIST) enriched the work with real-world testing insights.
This contribution could directly enhance user satisfaction on commercial search platforms, where minority intents affect millions daily. The paper's arXiv preprint details the full methodology, offering open access for fellow researchers.
Profile of a Prodigy: Rikiya Takehi's Journey
Takehi entered Waseda via the competitive Honjo International Scholarship Program, opting for the English-taught Information and Computer Science program in the Department of Computer Science and Communications Engineering. Initially undecided on graduate school, he dove into research during his third year, joining Professor Tetsuya Sakai's lab. A pivotal one-year leave of absence as a visiting researcher at NIST honed his skills in large language models and evaluation metrics.
His portfolio boasts first-author acceptances at fellow A* conferences SIGIR 2025 ("LLM-Assisted Relevance Assessments") and ICLR 2025 ("General Framework for Off-Policy Learning"). Internships at CyberAgent AI Lab and NVIDIA further polished his expertise. Takehi graduated in March 2026 as the recipient of Waseda's Ono Azusa Memorial Prize—the highest science and engineering honor, awarded to one in over 10,000 students—and multiple departmental accolades. Now interning at NVIDIA AI on agent systems, he eyes a U.S. PhD, with offers from top programs, aspiring to bridge Japan-U.S. AI advancements.
Mentorship at Sakai Lab: Catalyzing Undergrad Excellence
Professor Tetsuya Sakai's lab at Waseda specializes in information retrieval, natural language processing, machine learning, and human-computer interaction, with a social good emphasis. Sakai, a Distinguished Lecturer of ACM SIGIR, has guided Takehi since his junior year, crediting the student's rapid progress. The lab's culture—evident in recent wins like Noeko Fujii's IPSJ Yamashita SIG Research Award—encourages undergrads through hands-on projects, international collaborations, and conference submissions.
This model exemplifies how targeted mentorship transforms novices into conference stars. Takehi credits Sakai, Diaz, and NIST colleagues: "They provided invaluable opportunities and guidance." Such labs are vital in Japan, where undergrad research participation remains lower than in the U.S. but is surging at elite institutions.
Photo by Moughit Fawzi on Unsplash
WSDM 2026: The Gold Standard in Web AI
ACM WSDM is the premier venue for breakthroughs in web-scale data mining and search, attracting 1,000+ attendees annually. Past best papers have influenced Bing and Google algorithms. With topics spanning recommender systems to LLMs, the 2026 edition emphasized agentic models and ethical AI. Takehi's win as an undergrad—amid PhD-heavy submissions—marks a historic milestone, signaling Japan's rising IR talent.
Japan's Undergrad AI Research Renaissance
Historically, Japanese undergraduates focused on coursework, with research reserved for grad school. Yet, reforms like the Top Global University Project have boosted internationalization and hands-on opportunities. Waseda, ranked #293 globally for AI (EduRank 2026) and top-20 in Japan for CS (US News), exemplifies this shift. Stats show Japanese undergrads authoring 5-10% of A* conference papers, up from 2% a decade ago, driven by English programs and labs like Sakai's.
Challenges persist: funding lags U.S. peers, and cultural emphasis on conformity can stifle bold ideas. Successes like Takehi's inspire, with peers at UTokyo and Kyoto U securing NeurIPS spots.
Waseda's AI Ascendancy in Japan
Waseda boasts robust AI infrastructure, including the AI Center and English undergrad tracks drawing global talent. In QS 2026 Data Science rankings, it places strongly among Japanese peers, behind UTokyo but ahead in employability. Alumni lead at Sony AI and Rakuten, while initiatives like Waseda Research Awards fund breakthroughs. Takehi's feat elevates its profile, attracting recruits amid Japan's ¥10 trillion AI investment by 2030.
Global Implications for Search and Beyond
Takehi's VRisker re-ranker promises fairer search, vital as AI agents personalize results. In Japan, where LINE and Yahoo dominate, adoption could boost user trust. Broader, it advances risk-sensitive IR, applicable to recommendations and ads. As LLMs integrate, such math-grounded methods ensure equity.
Photo by Myznik Egor on Unsplash
Challenges and Future for Japanese AI Talent
Japan produces 10% of global AI papers but trails in undergrad output. Takehi's path—internships, leaves, collaborations—offers a blueprint. Waseda expands such via IPS grad school ties. Government Moonshot programs target 1,000 AI PhDs yearly. Takehi plans U.S. PhD return, embodying brain circulation.
Lessons for Aspiring AI Researchers in Japan
- Join labs early: Seek IR/ML groups like Sakai's.
- Leverage international stays: NIST/CMU boosted Takehi.
- Target A* venues: SIGIR/ICLR build momentum.
- Balance internships/research: NVIDIA honed agents.
- Persist: Takehi pivoted twice.
Waseda's ecosystem—English programs, prizes—lowers barriers. Explore research positions or Japan academic jobs.
