To create a project, complete the following steps.
-
On the project navigator, click the Add a project icon
.
-
In the Project Name field, enter the name of the
project.
- Select the Use positional information for classification check box, if you want positional information to be taken into consideration when learning the project.
- Select the Use UTF-8 encoding check box if you want to pass your documents in the UTF-8 format.
- Select the Use Classification Parameters check
box to activate classification configuration based on the given threshold
and distance. During classification, the system looks at the classes that
exist in the ALM project and uses a fuzzy document compare technology to see
if any of them are a match for the document being processed. The extent of
the match is measured by a confidence score where 100 is a perfect match and
0 is no match at all. The ALM class with the highest confidence score – the
best class - is used as the classification result, but you can refine this
decision process using the following parameters:
- The threshold (range 0-100, default 75) is the minimum confidence score that the best class must reach for classification to be considered successful. If the best class does not reach this threshold, then classification is not considered successful and no ALM processing results are returned. Only positive integers are allowed..
- The distance parameter (range 0-100, default 0) controls the minimum permitted gap between the confidence score for the best ALM class and the second-best class. You can use this parameter to stop the system classifying documents to the best class where other classes exist with a similar confidence. In this case, no ALM processing result is returned. The distance check is only carried out if the best class has already passed the threshold check. Only positive integers are allowed.
- Click Create.