What's required to do the use cases

Find Stuff

1. Search by strings including regular expressions
2. Proximity searches
3. Search by metadata

Analysis proper

1. Shallow parsing
2. Chi Sqare, Dunning's Log Likelihood ratio and similar routines
3. Text classifiers (Bayes, SVM, etc)
4. Statistical procedures for dimension reduction and clustering analysis

Visualization

1. each result set needs a visualization thus result sets need to be determined and visualizations could be reused as needed
2. link-node visualization for entity relationships
3. dendogram for clustering
4. timeline (http://simile.mit.edu/timeline/)
5. clustering of frequent patterns