[Pycon] [new paper] "Tariq Rashid" - A Plain English Overview of Text Mining Techniques

[info a pycon.it]

Title: A Plain English Overview of Text Mining Techniques
Duration: 60 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: Text mining and natural language processing (NLP) are currently experiencing a huge amount of interest. 

New techniques are being announced regularly and the ability to extract insight from natural language grows ever more impressive.

Despite the explosion of popular tools and powerful libraries, many of the ideas and methods are not widely understood, and not well explained for non-experts. 

How are two words similar? What is a word vector?  How can king - man + women = queen? Why would a neural network help? What is search result relevance? How does clustering work? How can complex data be visualised on a simple 2-d picture? Can we really extract distinct topics from a body of text? How can code write fairly realistic text?

This talk aims to demystify modern methods and explain the key ideas using plain English, lots of pictures, and a very minimum of mathematical jargon. 

If you're a user of Python tools such as NLTK or Gensim, or just want an intuition on recent advances, this talk is for you.


Tags: [u'Text-Mining', u'datamining', u'natural-language-processing', u'machine-learning', u'pydata']