A Structured Corpus for Physics Formula Retrieval with Prerequisite Dependency Mappings and Problem Decompositions
Description
This dataset contains 10,000 annotated physics word problems developed for formula retrieval, prerequisite-aware reasoning, physics problem solving, educational NLP, intelligent tutoring systems, and LLM evaluation. All problems, formulas, and conceptual content were sourced from the official SSC Physics 2026 English Version textbook published by the National Curriculum and Textbook Board (NCTB), Bangladesh. The source material was manually curated into structured problem-solving instances. Each record contains a physics word problem, primary formula, prerequisite concepts, and supporting formulas covering intermediate steps, unit conversions, variable definitions, and numerical substitutions. The dataset is distributed in UTF-8 JSON Lines (.jsonl): full.jsonl (10,000 instances), train.jsonl (80%), valid.jsonl (10%), test.jsonl (10%), and sample.jsonl (20 instances). Each record includes: id, chapter, topic, problem, prerequisites, primary_formula, and supporting_formulas. Example Record: { "id": "phy_009196", "chapter": "Radioactivity and Electronics", "topic": "Atomic Structure and Nucleus", "problem": "How much energy in Joules is released when 1.0 gram (1.0 x 10^-3 kg) of matter is entirely converted to energy? (c = 3.0 x 10^8 m/s).", "prerequisites": [ "Mass-energy equivalence principle", "Mass unit conversion to kg" ], "primary_formula": "E = m * c^2", "supporting_formulas": [ "m = 1.0 x 10^-3 kg", "E = (1.0 x 10^-3) * (3.0 x 10^8)^2 = 9.0 x 10^13 J" ] } Records were reviewed for completeness, annotation consistency, factual accuracy, and valid JSON formatting before release. The dataset supports controlled studies of formula selection, prerequisite dependency modeling, and structured problem decomposition across secondary school physics topics. This dataset is intended for research and educational purposes, particularly for building formula-grounded, prerequisite-aware, and curriculum-aligned physics problem solving systems.
Files
Steps to reproduce
1.Collect physics word problems from the SSC Physics 2026 English Version NCTB textbook. 2.Identify and annotate the primary formula required to solve each problem. 3.Map prerequisite concepts necessary for understanding and applying the primary formula. 4.Decompose each problem into 2–5 supporting formulas covering intermediate steps, unit conversions, variable definitions, and numerical substitutions. 5.Validate annotations through consistency checking, factual accuracy review, and cross-verification against textbook solutions. 6.Structure the dataset in UTF-8 encoded JSON Lines (.jsonl) format with stratified train, validation, and test splits.
Institutions
- American International University-BangladeshDhaka Division, Dhaka