I developed roberta-base-namecalling, a transformer neural network fine-tuned on ~12K social media posts annotated for the presence or absence of namecalling. This was joint work with my colleagues Steve Rains, Yotam Shmargad, Kate Kenski, and Kevin Coe.

TimeNorm: time expression normalization

I developed timenorm, a parser for normalizing time expressions initially based on synchronous context free grammars and later based on neural network models. It can recognize, for example, that if today is 2020-12-11, then “the next three weeks” should be normalized to the interval [2020-12-12T00:00, 2021-01-02T00:00).

ClearTK: machine learning in the UIMA framework

I was one of the lead developers for ClearTK, a UIMA based toolkit for developing natural language processing systems, which extends the UIMA API to make developing machine learning components easier.

uimaFIT: UIMA factories, injection and testing

We originally developed uimaFIT as part of ClearTK to provide general purpose utilities for working with UIMA components. It was subsequently adopted into the main UIMA distribution.

argparse: Python command line parser

I was the original developer of argparse, a command line parsing module for Python which is now part of the Python standard library.