Praline is an exact parser from a subset of natural language mathematics to an intermediate symbolic representation language based on logical forms, also called Praline. The Praline parser was implemented on a dataset derived from IMO shortlist geometry problem statements. The latest snapshot of the parser produces roughly 121 gold parses out of 578 total reference sentences, for a success rate of 21%.

Instructions

Building for testing and development

Note: These build instructions assume a unix-like system.

1. Create a workspace directory, and clone this repository into the workspace:

   mkdir praline-workspace
   cd praline-workspace
   git clone https://github.com/peterhj/praline
   

2. Run the bootstrap script in this repository to checkout the vendored dependencies into the workspace:

   cd praline
   ./bootstrap.sh
   

3. The default Makefile target is for a release build:

   make
   

   or, you may target a debug build:

   make debug
   

Running Praline on the IMO shortlist geometry reference set

The script for running the parser on the IMO shortlist geometry dataset is at tools/run_geom_r135.rs.

# release build:
./target/release/run_geom_r135

# debug build:
./target/debug/run_geom_r135

Datasets

The Praline parser was built using a reference dataset of transcribed IMO shortlist geometry problems. Additionally, the latest snapshot of the Praline parser produces a dataset of gold parses for a subset of the natural language sentences in the reference dataset.

IMO shortlist geometry reference dataset

The dataset consists of 578 sentences, in English LaTeX, transcribed from 135 geometry problem statements from the IMO shortlists of years 2006 to 2022 inclusive. The complete dataset in JSON Lines format can be found in this repository at the path data/geom_r135.jsonl.

Note: For pre-processing purposes, sentences in the dataset have been normalized to use single spaces, and also have a single space prepended.

Gold parse set

The set of complete gold parses produced by the Praline parser corresponding to the reference dataset are available at the path data/geom_r135_gold_latest.jsonl.

The latest snapshot of the parser produces 121 gold parses from of a total of 578 reference sentences, for a success rate of 21%.