David McIlroy

1) WHY IS TEXT ANALYSIS IMPORTANT

• TEXT IS THE RICHEST, MOST ACCESSIBLE STORAGE MEDIUM FOR KNOWLEDGE WE HAVE!
• Text is a "coded" storage medium - how to get the actual knowledge out of what others wrote down.
• The coding consists of:-

• Information overload. Needle in a haystack.
• Hidden patterns in text, e.g. trends, movements, changes in ideas and society.
• Critical review becoming impossible. Who is right?
• TACIT (gut feel) knowledge goes to EXPLICIT (knowledge you can write down) knowledge and back to TACIT. The most difficult code to unlock
• Language barriers.

2) WHAT WE ARE LOOKING AT.

Common theme is

Interactive

Visual

Appropriate analysis

Problem is

10 years worth of news paper / newswire reports = 5 to 15,000 articles.

Current Approaches are

A) GATE plus Network Visualisation (Analysts Notebook)

Gate is a text analysis program produced by University of Sheffield. The link to the department involved is

http://www.dcs.shef.ac.uk/research/groups/nlp/

The visualisation program used is Analysts Notebook from the company

http://www.i2.co.uk/
 
 

B) Themescape (SMART type algorithm for clustering plus 3D visualisation.)
 
 

This is a commercial program, site is

http://www.thememedia.com/