Diagram Understanding in Geometry Questions

Authors: Min Joon Seo, Hannaneh Hajishirzi, Ali Farhadi, Oren Etzioni

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To empirically evaluate our method, we compile a new dataset of geometry questions (textual descriptions and diagrams) and compare with baselines that utilize standard vision techniques. Our experimental evaluation shows an F1 boost of more than 17% in identifying visual elements and 25% in aligning visual elements with their textual descriptions.
Researcher Affiliation Collaboration 1University of Washington, 2Allen Institute for AI
Pseudocode Yes Figure 2: G-ALIGNER: Method for coupling primitive identification and alignment.
Open Source Code No The paper states 'Our dataset and a demo of G-ALIGNER are publicly available at:http://cs.washington.edu/research/ai/geometry'. This links to a project page and a demo, but does not explicitly provide direct access to the source code for the described methodology.
Open Datasets Yes The dataset is publicly available in our project web page. 1Our dataset and a demo of G-ALIGNER are publicly available at:http://cs.washington.edu/research/ai/geometry
Dataset Splits No The paper mentions evaluating against a 'ground truth dataset' but does not specify explicit training, validation, or test splits (e.g., percentages or sample counts for each partition).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU models, CPU types) used for running the experiments.
Software Dependencies No The paper mentions using tools like 'Hough transform' and an 'OCR package of Tesseract', but it does not specify any version numbers for these or other software dependencies.
Experiment Setup No The paper discusses 'parameters' for Hough transform but states their method is 'not sensitive to the choice of parameters' and does not provide specific hyperparameter values or detailed system-level training configurations for reproduction.