Loading…
Work through the research question, feasible methods, and fit with the target program. MIKO stages only evidence-bound proposals; you decide what to adopt, exclude, and confirm.
Many applicants turn the research plan into a book report — piling on passion and grand vision, yet never pinning down the research question, never engaging prior work, never offering a feasible method. That emptiness is spotted instantly at outreach, interviews, and faculty review, where it becomes the hardest barrier. A draft that holds up narrows the intent into one concrete question, then grounds it in the professor’s recent research direction.
Don't open from a topic area like "research AI." Force yourself to write one answerable question: who, under what conditions, answering what. Nail that sentence and the whole draft has a spine; leave it blank and the rest is just piling on.
Before you write, look up who has done what in this field, how far, and what's still missing — then place your question right in that gap. With a foothold, the plan reads as continuing the academic conversation rather than proposing from thin air.
Build it straight off the question: where the data comes from, sample size, the steps, the analysis. Squeeze it down to what finishes in a year or two — a plan is judged not on ambition but on "you can actually pull this off."
Once a lab is chosen, draw your topic and method from the professor's papers, KAKENHI projects, and keywords of recent years — not the stale lab blurb. The closer the fit, the more it reads as "this person could join my group."
Better to cite a few papers you've truly read than to pad the list from memory or imagination — fabricated citations don't survive a check. This tool gives format and search-keyword guidance only; you supply the references from fact, and it never generates them for you.
Each graduate school sets its own rules on length, style (だ・である register), and references. Framing your section outline to those requirements from the start beats rewriting the format after the prose is done.