Blog

Multilingual Data Stats

TLDR Multilingual Wikipedia is about 80GB.
...

XNLI, Cross-lingual Natural Language Inference

I didn't make it, but I tried to explain it.
...

The Machine Learning Model Library of Babel

In this library, all possible machine learning models exist.
...

2020 NLP Wish List

Things I would like to see happen in NLP in 2020.
...

Vocabulary in Allennlp

tldr; always include test in your config.
...

Can you use BERT to generate text?

No.
...

Entities are Important

Who cares about NER anyway?
...

Cross-lingual Transfer of Named Entity Recognizers without Parallel Corpora

I recently came across this paper: Cross-lingual Transfer of Named Entity Recognizers without Parallel Corpora, by Ayah Zirikly and Masato Hagiwara, published in IJCNLP, 2015. It’s very related to my research, so I wanted to write a short review.

...

Workshop on Balto-Slavic NLP

This year, the 6th Workshop on Balto-Slavic NLP was held, co-located with EACL 2017, in Valencia, Spain. The associated shared task was formally called Multilingual Named Entity Recognition, but in fact consisted of 3 subtasks:

...

Kids Books

I read a fair number of kids books in my line of work. Some are excellent, while others should never have been published. Here's a short record (updated sporadically).
...

Supervision

How many different types of supervision are there? I've collected as many as I could find, along with an example paper.
...

Lexical features compared to Brown Cluster features for NER

Lexical features vs brown clusters
...

How to programmatically train LBJava

You CAN teach an old classifier new tricks.
...

How to use illinois-ner with Maven

Here are some crazy tips UIUC researchers are freaking out about!
...

Remote Server Workflow

A note on how I set up and use a testing server.
...

How Many Languages Per Script?

Language stats
...

Weights in LBJava

How to get weights in LBJava
...

Illinois Transliteration

How to use illinois-transliteration
...

Retrofitting Retrofitting

My thoughts and experiments on the retrofitting paper.
...

Handglovery

This is my first post, please do not be alarmed.
...