mathipy

Multimodal item feature extraction for K-12 math assessment. Reads item text and images through one entry point, with math-aware normalization via textstat and NLTK. Measure readability, Common Core State Standards for Mathematics domain, cognitive load, and the relations carried by wording or by notation. Classify item images into visual types with an instructional role, read shape subtypes with OpenCV and Pillow, and extract text through Gemini and OpenAI vision APIs.

Installation

pip install mathipy

With optional dependencies:

pip install "mathipy[nlp]"        # readability (textstat, nltk)
pip install "mathipy[vision]"     # visual analysis (opencv, pillow)
pip install "mathipy[ocr]"        # OCR via vision LLMs (httpx)
pip install "mathipy[documents]"  # document parsing (python-docx, pdfplumber)
pip install "mathipy[all]"        # all features

Citation

• Shin, M. (2026). mathipy: Multimodal item feature extraction for K-12 math assessment (Python package version 0.4.2) [Computer software]. https://github.com/mshin77/mathipy