Thursday, June 21, 2012
Tuesday, June 12, 2012
Tech Review AI: Tuesday, June 12, 2012
- Have listen to the course introduction
- listening to the Basic Text Processing -> Regular Expressions
- Regular expressions
- a formal language for specifying text strings
- regular expression disjuctions
- letters inside square brackets []
- pipe disjunction
- a|b|c is the same as [abc]
- ranges [A-Z]
- negations [^A-Z]
- ?
- previous character is optional
- .
- one charater
- ^
- begining of line
- $
- end of line
- \.
- use backslach to find special characters
- Kleene operators
- Stephen C Kleene
- * is 0 or more
- + is 1 or more
- non-alphabetics
- regexpal.com
- Type1 error false positive
- increasing accuracy or precision (minimizing false positives)
- Type2 error false negatives
- increasing coverage or recall (minimizing false negatives)
- Type1 and 2 conflict, antagonistic
- regular expressions play a surprisingly large role in NLP.
- regular expression used as features in classifiers for the hard NLP tasks
- Basic Text Processing -> Regular expression in practical NLP
- Christopher Manning
- Stanford English Tokenizer
- a deterministic
- Java Flexer
- {2,4}
- length between 2 to 4
Subscribe to:
Comments (Atom)
