Supporting an Effective Text Analysis Methodology

To analyse free format text held within a set of messages. There are a number of steps that it is advisable to follow:

a) Define a TextProcessor Object for each domain.

b) Define the Attributes of the domain.

c) Scan a sample of messages, highlighting phrases and mapping them to Attributes.

d) Review the accumulated Phrases looking for synonyms and adding wild-cards.

e) Test accuracy of classification and false classifications (false positive).

f) Enhance expressions with further wild-cards and synonyms.

Define a TextProcessor Object for each Domain

Define the Domain Attributes (Object Phrases Tab)

Scan Sample Messages (Phrase Matching Tab)

Review Accumulated Phrases

Testing the Accuracy

Enhance Expressions with Further Wild-cards and Synonyms