go-v.orgdoi

10.63808/css.v2i3

— · دوفصلنامه راهبردهای شناختی در یادگیری · 2026

  • DOI 10.63808/css.v2i3
  • Publisher OpenAlex
  • Type paratext
  • Issued 2026

Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.

دوفصلنامه راهبردهای شناختی در یادگیری · OpenAlex · paratext · 2026

Abstract

General large language models suffer from factual hallucinations, mismatched expression difficulty and insufficient individual adaptation in science education question-and-answer (Q&A) tasks.To address these issues, this paper proposes a personalized popular science Q&A generation method integrating difficulty-aware retrieval-augmented generation.Firstly, a scientific knowledge base with academic stage difficulty labels is constructed in this method.Secondly, learner portrait matching items and knowledge difficulty deviation penalty items are introduced into the traditional semantic retrieval score to realize evidence screening for students in different academic stages.In the generation stage, structured prompts are organized with retrieved evidence, student portraits and answer constraints, and parameter-efficient fine-tuning is adopted to enhance the model's adaptability to popular science Q&A style and knowledge boundaries.Taking a self-built primary and secondary school science Q&A dataset as the research object, experiments are conducted from the dimensions of scientific accuracy, age appropriateness, completeness and hallucination rate.The experimental results show that the proposed method outperforms conventional retrieval-augmented methods in terms of scientific accuracy and age appropriateness, and can improve the reliability and teaching applicability of popular science Q&A with low training costs.

Open the resolved locator · 10.63808/css.v2i3

Metadata licence: OpenAlex metadata (CC0)

Registry
OpenAlex DOI index + API
Fetched
2026-11-20 18:25 UTC
Cache state
not cached yet (bulk-index+openalex)
Locator
https://system.wisacad-pub.com/index.php/css/login
Metadata licence
OpenAlex metadata (CC0)
Scope
Metadata as returned by the resolver; abstracts are never generated here.