Software Engineer (contractor) at Stealth Startup
I've worked at several decent companies in Silicon Valley:
- Databricks (a company that maintains Apache Spark) - worked on backend & infrastructure in Scala, but I've made a couple of commits to Spark too.
- LinkedIn - I worked on candidate reccomendations for recruiters.
- Facebook - statistics for Ads Manager (plz don't hate me for working on ads).
- Minted - web-dev in python.
* Currently working in a stealth startup on NLP tasks. For the last 2 month I was building a text categorization system from scratch, so I wanted to share my experience and findings from that.
* Education-wise, I did my undergrad at University of Waterloo. Throughout my studies I did a lot of competitive programming, reaching ACM ICPC finals (and placing 13th there) and becoming red on TopCoder.
Topic of presentation: Overview of text classification approaches: algorithms & software
The main points of the presentation:Overview of text classification approaches: algorithms & software
Summary: For the last 2 month I've been building a system for classifying customer support tickets into several categories in terms of product area, importance, etc. Throughout that time I've tried several approaches and benchmarked them against each other. In this talk I would like to showcase some of my findings, including research algorithms that perform well and relevant software. This talk would be useful for someone who needs to build a text categorization system, or someone who just wants to get an overview of one of the most popular NLP research problems (classification).
In this talk you will learn:
* About various approaches used for text classification (e.g. approaches based on TF-IDF, or approaches based on word embeddings and RNNs - recurrent neural nets).
* How these approaches perform against each other on a real-world data.
* Software that is useful for implementing these approaches.
* Research behind some of these approaches.