Using Custom extraction rule in text analysis is suppressing the analysis process.

[Former Member]

Following this blog post http://scn.sap.com/community/developer-center/hana/blog/2014/05/23/to-be-or-not-to-be-hana-text-anal... , I have tried to create a table with a text index with a custom configuration all is working fine until I reference a custom extraction rule file (CUSTOM.fsm) which seems to totally disable the analysis process i.e. when I insert new text in the table there are no analysis results in the $TA_table when I remove the reference from the config file the analysis process returns to a working state.

I cannot understand what is happening I checked against the SAP HANA Text Analysis developper guide and the text analysis customization guide but cannot find the problem.

Is there a way to debug the analysis process?

Should I place the extraction rule files in specific place?

[Former Member · Accepted solution]

In sap hana the good extension for the custom text rule is .hdbtextrule and not .fsm.

By correcting the extension I was able to get analysis result.

[Former Member]

This message was moderated.

[martin_chambers]

Hi Chedy,

did you ever find out how to debug text analysis?

I have created a customer dictionary and added a line in the config file.

My config file still works, but the dictionary file is completely ignored and there is no error message.

Do you have an idea how I could analyze this situation?

Regards,

Martin

[Former Member]

Hello Chedy !

Where did you find out how to write custom extraction rules for text analysis ?

Is there any blog or pdf file where I can learn how to create custom rules ?

Thanks,

Christoforos

[Former Member]

Hi Christoforos,

You can find guidance about writing custom rules in this document:

SAP HANA Text Analysis Extraction Customization Guide - SAP Library

Regards,

Chedy

[Former Member]

Thank you for your quick reply !

So do you think that a full greek dictionary should be made ?

Is this the only way ?

I was wondering whether there is a more efficient and automated way to do it !

Christoforos

[Former Member]

It depends on what you intend to do with text analysis.

Try to find out if Greek is supported (many languages are), if so I think there is no need to create a full Greek Dictionary if you only intend to do basic entity extraction.

[Former Member]

I would like to do EXTRACTION_VOICEOFCUSTOMER, because it is not supported yet.

I used that PAL function on Greek text, but I got no results back - so I had to translate into English and then do the function.