Here are some text mining project ideas from Yi Zhang.

1.	Sentimental text classification: classifying review documents according
to the polarity of its prevailing opinion (favorable or unfavorable).
(corpus: epinion review data, text classification)
2.	 Email classification: automatically classify emails into different
folders of each person (corpus: Enron email, hierarchical text
classification)
3.	Study how to choose text classification algorithms (feature selection,
feature numbers, classification algorithms) for different data set.
(corpus: several data set with very different characteristics and class
level)
4.	Topic detection and tracking: discovering and threading together
topically related material in streams of data such as newswire and
broadcast news (online text clustering)
5.	  Assign Gene Ontology (GO) to full text documents from two years of
three MEDLINE journals. (Text classification)
6.    Favorite page recommendation (text classification)
7.    Extract disease and treatment from bioscience literature (entity
classification using classification algorithms such as NN, or information
extraction algorithm such as HMM)
8.	   Using web to improve text classification accuracy

Except task 6, I have the corpus and the right answer for students to
evaluate the algorithms. Hope this list is useful for your class. Please
let me know if you need more information.

Regards,

Yi