MoNoise

MoNoise is currently the state-of-the-art model for lexical normalization for multiple datasets. Lexical normalization is the task of translating social media texts to its canonical equivalent.

Demo page

Unrestricted analogies

Analogies such as man is to king as woman is to X are often used to illustrate the amazing power of word embeddings. Concurrently, they have also exposed how strongly human biases are encoded in vector spaces built on natural language. However, mainstream implementation often ignore the input words to find the most likely X. In practice, including the input words (`man', `king' and `woman') often leads to returning the second word. In this demo we removed all constraints from standard 3cosadd implementation.

Demo page