AIIRLab/MathDisAM

1---2configs:3- config_name: default4  data_files:5  - split: train6    path: "**/*train*.tsv"7  - split: test8    path: "**/*test*.tsv"9size_categories:10- 1K<n<10K11---12 13# MathDisAm: Towards Math Query Disambiguation14 15Welcome to the official repository for the paper **MathDisAm: Towards Math Query Disambiguation**. This repository contains the dataset, query generation pipelines, and baseline classifier models designed to distinguish between ambiguous and unambiguous mathematical queries.16 17## Dataset18 19The `data` directory contains the core datasets used for training and evaluating the disambiguation models:20* **`MathDisAM_train.tsv`**: The training file containing MathSE IDs, mathematical questions, and their corresponding tags ("Ambiguous" or "Unambiguous").21* **`MathDisAM_test.tsv`**: The untouched, held-out test file used for final model evaluation.22 23---24 25## Query Generation 26The query generation process leverages a two-agent Large Language Model pipeline (using Gemma) to process natural mathematical questions:27 28* **Ambiguous Query Generation**: 29  * Uses an "Injector" agent to rewrite a math question to introduce deliberate ambiguity. 30  * Uses a "Verifier" agent to independently judge if the rewrite is genuinely ambiguous.31  * The script processes questions until a target of 1,000 verified ambiguous questions is produced.32* **Unambiguous Query Generation**: 33  * Uses a "Clarifier" agent to rewrite queries to remove ambiguity. 34  * Uses a "Verifier" agent to evaluate if the query is fully well-posed and unambiguous. 35  * This pipeline also targets 1,000 verified unambiguous questions.36 37---