Google has deprecated legacy versions of AutoML services, which directly impacts IDP's core functionality. Additionally, the IDP application was deprecated with Appian 23.2. Customers who wish to use the application will need to refactor plug-ins using AutoML. |
Your Appian installation likely includes different environments, such as development, staging, and production.
When you install Intelligent Document Processing (IDP) or update the pre-installed application, you will likely be doing this in a source environment, such as development. You will also likely use this source environment to fine-tune your application before deploying it to a target environment, such as staging or production.
It is best practice for each environment to use its own classification model and data extraction training. Because of this, in order to deploy changes to a target environment, you will have to take certain steps.
This page provides instructions for deploying changes to target environments, while making sure that each environment contains its own classification model and data extraction training.
See also: The Deploying Apps pages in the Appian Designer documentation for more information about deploying patches in Appian.
If you are using IDP in a subprocess or with external systems, then you are likely also deploying a new application or deploying a patch for an existing application. In this scenario, it is recommended that you first complete the deployment for IDP and then complete the deployment for the dependent application or patch.
After you have installed or updated your pre-installed application in your source environment, customize the application using the resources mentioned in the Customizing IDP section of the Intelligent Document Processing Overview page.
Before you continue with deploying changes to a target environment, make sure that you have followed the instructions in the Installing IDP page to either install IDP or update the pre-installed application in your target environment.
Tip: If you are using Compare and Deploy, you may deploy the patch and the associated SQL at the same time.
If you add or remove a field from a document type in your source environment, to update the document type in the target environment, you will need to make the following changes:
The updated CDT will only display in new process instances. It will not display in any process instances that were started before you deployed the modified document type.
If you add a new document type in your source environment, it is best practice to use a new classification model in your target environment. This is because you will likely be uploading test documents in your source environment, and you don't want the classification model for your real world documents to be affected by test documents.
In order to use a new classification model in the target environment, you will need to set the status for the new document to Inactive
. This is because you will need to run the IDP configuration in the target environment to train the classification model. During the configuration, the status wil be updated to Active
.
To deploy a new document type to a target environment:
dudoctype
table, export the data for the new document type. This will generate the INSERT commands to add the document types to the table in the target environment. Alternatively, you can write your own SQL commands to insert the new rows into the table.doctypestatus
to Inactive
.
See the following for an example SQL command. Note that this example uses MySQL syntax.
1
INSERT INTO `dudoctype` (`doctypeid`, `doctypename`, `choiceindex`, `doctypestatus`, `isinvalidtype`, `channelid`) VALUES (NULL, 'Medical Device License', '5', 'Inactive', '0', '1');
dudoctype
table of your target database.DU Data Store
.DU_returnDataTypeForChoiceIndex
rule.DU_returnDataStoreEntityForChoiceIndex
expression rule.After the model is trained, you can start uploading documents for the new document type.
If you add a new document channel in your source environment, it is best practice to use a new classification model in your target environment. This is because you will likely be uploading test documents in your source environment, and you don't want the classification model for your real-world documents to be affected by test documents.
In order to use a new classification model in the target environment, you will need to reset the model training values for the new document channel in the target environment. You will also have to deploy the new document types that were created for the new document channel and train the classification model.
To deploy a new document channel to a target environment:
dudocchannel
table, export the data for the new document channel. This will generate the INSERT commands to add the document channel to the table in the target environment. Alternatively, you can write your own SQL commands to insert the new rows into the table.modelid
: NULLmodeltrainedon
: NULLnumdocsusedfortraining
: 0invalidtypeincludedinmodel
: 0Example:
1
INSERT INTO `dudocchannel` (`channelid`, `channelname`, `modelid`, `modeltrainedon`, `minclassifierconf`, `minextractionconf`, `numdocsforretraining`, `numdocsusedfortraining`, `invalidtypeincludedinmodel`) VALUES (2, 'Legal', NULL, NULL, 0.85, 0.5, 100, 0, 0);
dudocchannel
table of your target database.Deploying Patches