Question
IKOR
AU
Last activity: 15 Sep 2017 3:40 EDT
Text Classification Model jar file is not deploying in Pega NLP.
I have created a Text Analyzer rule and trying to Taxonomy rule (Decision Data) for text classification Using Pega NLP. It seems like server fails to generate the model file for some reason. This happens only in development server and in local environment setup its working as expected. I have added few screens shots.
1. Used the add language wizard and selected the language as English and Build new Model option to start the Model Development wizard.
2. Taxonomy and classification files are configured correctly and identifies with correct values.
3. After the Model Analysis and Modal selection steps when I try to download the generated model file it doesn't create the jar file but shows a activity status good window.
4. When I open the created model file from rule explorer it doesn't contain a jar file.Tried to manually upload a generated jar of a local server but still the text analyzation doesn't return any response.
5. In the log files NLPResourceAccessor is giving an error when trying for the model creation wizard.
Not sure whether this is related to any access restriction for model/jar creation in dev server.This is a new setup of a Pega 7.3 server.
Did anyone had similar problems while trying this kind of a scenario?
IKOR
AU
In activity PegaAccel-Management-Decision-NLP-CreateModel.pzGetModelProcessStatus step 1 java execution, pyWorkPage.pyModelDetails.pyAlgorithmList(1).pyFileSource is empty . it contains score , results and accuracy score.
In activity PegaAccel-Management-Decision-NLP-CreateModel.pzGetModelProcessStatus step 1 java execution, pyWorkPage.pyModelDetails.pyAlgorithmList(1).pyFileSource is empty . it contains score , results and accuracy score.
Pegasystems Inc.
US
IKOR
AU
Hi Gudis ,
I have tried with changing the algorithms but still it failed to attach and generate the file source. I see some log errors related to NLP generation after executing PegaAccel-Management-Decision-NLP-CreateModel.pzGetModelProcessStatus java step 1(see the attached image 5.png of original post).can you share me a FTP link to send the log file related to it.
*****Edited*****P.S.******
Log errors were related to csv file and if I select the Taxonomy rule from dropdown mentioned log errors didn't occur .
Pegasystems Inc.
IN
Hello!
I had a quick check with our SME, can you confirm if the model is generated after selecting the right taxonomy?
Regarding sharing the log file. The original post as you know, supports file attachments; the Replies also support file attachments. Are you able to attach the log file right here? If yes, make sure that you remove any proprietary information before posting. Else, we suggest you create an SR with Pega Support so that someone can look into it for you!
Regards,
IKOR
AU
No not able to generate the .jar model files and there are no exceptions in the log.
In the activity pzGetModelProcessStatus Step 1 java line 61 -> algorithm.putString("pyFileSource",Base64Util.encodeToString(mlOPResultBean.getModelBytes())); This returns as null.
The file source byte stream for the algorithm result is not generated.
Pegasystems Inc.
IN
Hello!
I had another quick check with our SME. We recommend that you create an SR with Pega Global Customer Support as this needs us to look into the system to find out what settings are causing the model to not generate.
Please let me know the SR number once created so that I can tag the two and add the engineer in loop with the SME. Also please add the URL to this post while creating the SR.
Regards,
IKOR
AU
Moved forward with it a temporary solution by generating the model binary stream from local server and set it on the dev server code when it set the binary file source .
It would be great if there is a permanent fix for this .