Chain-of-Concept: topic-graph organisation of past assessments replacing similarity-based retrieval¶

Design · All designs
Evidence · no claims cited

Description¶

Chain-of-Concept is the article's framework for curriculum-grounded retrieval. Rather than storing past assessment records sequentially and relying on surface-level embedding similarity, it organises records as a topic graph in which questions are decomposed into sub-question nodes labelled with curriculum topics, and retrieval is confined to the topic subgraph matching the query's topics. The authors state that topic filtering removes cross-topic noise, yielding more coherent evidence, and that the module then analyses pupil errors and prior knowledge to choose the most effective feedback strategy.

Design Implications¶

Context¶

Requirements¶

Questions must be decomposed into sub-question nodes and labelled with one or more curriculum topics
A FAISS-based ranker is used for scoring within the filtered topic subgraph

Constraints¶

Retrieval is confined to the topic subgraph matching the query's tagged topics, so records sharing no topic with the query are excluded

Target Learners¶

GCSE science students

Learning Goals¶

Personalised, curriculum-aligned feedback generation

Claims¶

Examples¶

-

Key Sources¶

Runcong Zhao, Artem Bobrov, Jiazheng Li, Cesare Aloisi, Yulan He. (2025). LearnLens: LLM-Enabled Personalised, Curriculum-Grounded Feedback with Educators in the Loop. https://arxiv.org/abs/2507.04295

Page metadata

Field	Value
`type`	`design`
`id`	`chain-of-concept-topic-graph-memory`
`title`	`Chain-of-Concept: topic-graph organisation of past assessments replacing similarity-based retrieval`
`description`	`Chain-of-Concept is the article's framework for curriculum-grounded retrieval.`
`status`	`review`
`generated`	by `process:settle-candidates` on `2026-10-12`
`trust`	unverified — no one has confirmed this page's content
`sources`	1 — `runcong-zhao-2025`

Frontmatter is the page's machine-readable record — see the Schema & Guide.