duplicate column error from weka predictor... knime predictors work fine

Ok I have used the text processing example to create a workflow that processing some documents, extracts keywords, partitions the dataset and then learns a classifier to classify documents in to categories.  It is exactly like the example workflow that comes in the workbench.  I can use the knime svm learner with knime svm predictor-- everything works fine and dandy.

However, I want to try some of wekas multinominal classifiers.  So I tried just putting a weka classifier learner and the general weka predictor.  Then I placed a scorer after the weka predictor.  No matter what I do I always get an error out of the weka predictor like: 

ERROR     7)                                 Configure failed (IllegalArgumentException): Duplicate column name "acq" at positions 7668 and 8214.
ERROR     7)                                 Configure failed (IllegalArgumentException): Duplicate column name "acq" at positions 7668 and 8214.
ERROR     7)                                 Configure failed (IllegalArgumentException): Duplicate column name "acq" at positions 7668 and 8214.


acq is one of the document classes and I dont think its even a column... but i suppose it could be if a stemmed keyword happens to be "acq" as well. 

Anyone have any ideas what is going on here?  As I'm typing this I'm wondering if maybe the predictor is trying to output a column for each class (for some reason) and not putting any pre/suffix so its colliding with the keywork.  Maybe ill try doing a string manip column to prefix all of the class names with CLS_ or something to make sure theres no collisioin and see if that is a good workaround.


For anyone else that stumbles across this -- that was the problem.  I just used a string expression joiner to add CLS_ to the beginning of the class values in the class column and it was fine.  the weka predictor adds columns for each of the classes (for some reason) and so that was conflicting with the column for the keywords

Hi Team,

I started working on knime and its really nice and easy to use.

i started one big workflow and finally i am going to handover this flow to my client. 

However i am facing few issue in knime as below.

1. i am trying to convert Value  Filepath to a Variable using "Value selection Quickform" under quickform node, and am feeding it to variable based file reader to ready file to read file in from the input path. but when i do this, for the first time, Output from "Value selection Quickform" reflects the correct value as input. However in next iteration, if  the input fvalue(Filepath) changes, output of "Value selection Quickform" still throw Previous Value(filepath). Which means the output of the same is not dyanamically changes the input changes. 

Please let me know asap since i have to release this to my client. 

and also please let me know if there are any nodes which convert input value to variable dyanamically as input changes. so that i can use variable based file reader.

2. In my work flow i have many CSV output to track the data. However when i give this entire workflow to my client, the Output path has to be configured once again for individual CSV output. Is there any way to use flow variable to write CSV as output. so that i need not configure all CSV output as per clients path.

3. When i transfer entire workflow to some other desktop, XLS reader will be correpted. It will fail to read the file if at all we browse file from correct location. At this point of time we wll have to delete current XLS reader an will have to use new XLS reader.

I appreciate your support. thanks a lot for your help



Shrinivas Nayak