Join NYC Python for a night of talks about loosely and dynamically typed languages with no discernible syntax rules!
Slicing Up Words with Python - Steven Butler
Morphological segmentation is the process of splitting words up in their component morphemes (smallest grammatical units in a language*) in order to improve a variety of downstream NLP tasks, like machine translation and keyword search. For well-resourced languages with well-understood morphological patterns (like English), this is no longer very much of an issue, but for low-resource languages (those that don't have a lot of NLP research done on them), this problem is a lot harder. I'll demonstrate a Python library called Morfessor (an interface to a family of unsupervised segmentation algorithms developed by a research team in Finland) that has been used for working on low-resource language segmentation, and talk a little bit about my work, which is trying to improve segmentation accuracy when morphological patterns get exotic**.
* e.g., 'words' -> ['word', 's']
** infixation, partial reduplication, etc.
Jargon-term Extraction by Chunking - Angus Grieve-Smith, Adam Meyers
One major subfield of NLP, Information Extraction (IE), deals with finding needles in haystacks - taking a huge corpus of text and pulling out the most important parts, like frequently used entities (people, places or things) and the relations between them. At NYU we developed methods for extracting jargon terms - terms that are used more frequently in particular fields than in general writing. These methods involved a combination of close annotation of the documents, algorithms to find particular patterns, and machine learning to extend the annotation.
N-grams: What they are and what you can do with them - Max Schwartz
I will touch on text classification, text generation, and how N-grams can fit in to solutions to problems like word segmentation and part-of-speech tagging.
The Ministry of Silly Talks
About our Sponsor:
Based in New York City, Vettery is a trusted hiring marketplace built to help tech candidates find their dream job. They take the best from the recruiting and technology worlds by connecting employers and talent through a curated marketplace of vetted candidates. Vettery currently works with over 1,000 companies including Uber, Jet, Etsy, Flatiron Health, Digital Ocean, SeatGeek, and Casper. For more information on being featured, please apply on vettery.com.