Oracle Data Mining Application Developer's Guide 10g Release 1 (10.1) Part Number B10699-01 |
|
|
View PDF |
The PL/SQL interface enables you to perform Text Mining using a simple two-step process:
Step 1: Given a Text document table, and an Oracle Text Index built against the documents, extract the text "features" using a simple PL/SQL driver provided with the ODM installation (see Section 5.2, "Summary of ODM PL/SQL Sample Programs"). This driver demonstrates how to store all the text features corresponding to a DocID into a table with nested table columns.
Step 2: Provided the table created in Step 1 as input to the CREATE_MODEL
or APPLY
operation, as appropriate, to be classified using any classification algorithm -- such as SVM, or a clustering algorithm such as k-Means, or a feature extraction algorithm such as NMF.
Note that this two-step process is flexible and can handle any general text input; you just have to provide the text features in an input table whose schema corresponds to the one depicted in dm/demo/sample/plsql/textfe.sql
.
See also Chapter 13, "Text Mining Using ODM," in Oracle Data Mining Concepts.