Home

Welcome to the am-parser wiki!

This documentation is a work in progress. If anything is unclear, please open an issue and we will get back to you as quickly as possible.

A characteristic feature of the AM parser is that it learns to parse sentences into AM dependency trees, which then evaluate to graphs in the AM algebra. We have defined our own file format, AM-CoNLL, to store AM dependency trees; it is a mild extension of the well-known CoNLL format for storing dependency trees.

Training the AM parser

Training the AM parser consists of two steps:

1. Preprocessing and decomposition: convert the graphbank to AM dependency trees
2. Training the neural network to predict AM dependency trees

Running the AM parser

Directly parsing sentences into graphs (with our pretrained models or the one you trained above) is explained in the main README.

Alternatively, you can use some other parser to produce trees in the AM-CoNLL format, e.g. the A* parser in the am-tools repository, and just let the am-parser compute the scores. In this case, you will need to evaluate the AM-CoNLL trees to graphs yourself.

Smatch-based evaluation for EDS and AMR requires some extra steps, which are described here.

Utilities

• Visualizing AM dependency trees
• Visualizing graphbanks
• UdS specific documentation