Configuring cognitive service for a custom application by using BMC Native (Google) classification

The following image illustrates the end-to-end process of configuring the BMC Native (Google) classification service in BMC Helix Innovation Studio. It describes the steps required and the UI tools used to configure BMC Native (Google) classification service.

Setting up the Google Cloud Platform account

You can set up the Google Cloud Platform accounts in one of the following ways:

• To set up the Google Cloud Platform account manually
• To set up the Google Cloud Platform account by using an onboarding script

To set up the Google Cloud Platform account manually

If you are using Google Cloud Translation Service for real-time translation, use this method to set up the Google Cloud Platform account.

1. Log in to Google Cloud Platform Console.

2. Create a new Google Cloud Platform project.

   1. On the Dashboard, click Create.

   2. On the New Project page, enter the following details:

      1. In Project Name, provide a meaningful name for the project.
      2. In Location, provide a region where the data sets will be trained and deployed.

      Important

      Project ID is generated by the Google Cloud Platform. Project ID is the globally unique identifier for your project. You cannot change the Project ID after the project is created. 

      Make sure that you note this Project ID generated by the Google Cloud Platform. You will need this Project ID while configuring the GCP-Project-Id setting on the Centralized Configuration UI in BMC Helix Innovation Studio. For more information, see Configuration settings.

3. Enable billing for this project.

   1. From the left navigation menu, select Billing.

   2. Click Link a billing account.

      Important

      If this is not your first billing account, make sure that you open the billing account list by clicking the name of your existing billing account near the top of the page, and then clicking Manage billing accounts.

   3. Enter the name of the billing account and your billing information. The options you see depend on the country of your billing address. 

      Important

      For United States accounts, you cannot change tax status after the account is created.

   4. Click Submit and enable billing.
4. Create a service account for your project.

   1. Select the project and click Service accounts. 
   2. On the Service accounts page, click CREATE SERVICE ACCOUNT.
   3. On the Create service account page, enter the following details:
      1. In Service account details, provide a meaningful name for the service account. For example, bmc-gcp-serviceaccount.
      2. In Service account ID, an Id is auto-generated by the Google Cloud Platform.
      3. In Service account description, provide a meaningful description of what this account will do.
      4. Click CREATE.
   4. On the Grant this service account access to project tab, provide the following permissions to the service account.

      1. To add all the roles, click ADD ANOTHER ROLE and select the following roles:
         
         

         Purpose	Role Category	Name of the role
         Google classification	Project	Owner
         	Storage	Storage Admin
         	Machine Learning Engine	ML EngineAdmin
         Real-time translation	Cloud Translation	Cloud Translation API Admin
         	Cloud Translation	Cloud Translation API User

   5. Click Continue. 
   6. On the Grant users access to this service account tab, grant access to users and generate the Create Key.

      1. Click CREATE KEY.

         Warning

         Store the file securely because this key cannot be recovered if lost.

      2. To select a download format for the key, select JSON, and click Create. 
      3. Rename the .json file as <projectID>-key.json.
   7. Click Done.
5. Create a bucket.

   This step is required only for BMC Native (Google) Classification. You can skip this step for enabling Google Cloud Translation Service.

   1. On Dashboard, click Cloud Storage.
   2. Click Create bucket to open the bucket creation form, and enter the following details:

      1. In Name, provide a meaningful name for the bucket. 

         Important

         Make sure that the name of your bucket must be the same as your project ID.

      2. Select Regional for Storage class and for Location make sure that your Google Cloud Platform machine learning job is able to access it. For more information, see region values on https://cloud.google.com/ml-engine/docs/tensorflow/regions.

      3. In Access control model, select Set object-level and bucket-level permissions.
      4. Click Create.
6. Create folders inside the bucket.

   This step is required only for BMC Native (Google) Classification. You can skip this step for enabling Google Cloud Translation Service.

   Click Create Folder to create the following metadata files: 

   Folder	Folder content
   credential	Content from the service account credential .json file <projectID>-key.json created in step 4.
   data	empty folder
   model-code	Upload the sequential-tokenizer-0.1.tar.gz file. For information about the file, see PDFs, videos, and API documentation.
   job-dir	empty folder
   results	empty folder

7. Enable the APIs and Service.

   1. From the Navigation menu, select APIs & Services.
   2. On the APIs & Services page, click ENABLE APIS AND SERVICES.

   3. Search for the following APIs and Services, and click Enable. 
      
      

      Purpose	API name
      Google classification	Cloud Machine Learning Engine (ml.googleapis.com) Google Cloud Storage JSON API (storage-api.googleapis.com) Identity and Access Management (IAM) API (iam.googleapis.com) Compute Engine API (compute.googleapis.com) Cloud Resource Manager API (cloudresourcemanager.googleapis.com)
      Real-time translation	Cloud Translation API (translate.googleapis.com)

   
   

To set up the Google Cloud Platform account by using an onboarding script

1. Log in to Google Cloud Platform Console.

2. Install Google Cloud Platform SDK.
   For more details about Google Cloud Platform SDK and installing it, see https://cloud.google.com/sdk/.

3. Run the following commands to make sure that gcloud and gsutil are installed in the bash shell.

   gcloud --help
   gsutil --help
   
4. Download a local copy of the following machine learning job code's compressed files:
   • sequential-tokenizer-0.1.tar.gz
   • gcp_onboarding.sh
     

5. Change the execution mode for gcp_onboarding.sh by using the following command:

   chmod +x gcp_onboarding.sh

6. (Windows only) Create bash functions so that gcp_onboarding.sh can use these functions. Use the following commands and add the functions with correct paths into the user's .bashrc file.

   gsutil() {
   /c/Program\ Files\ \(x86\)/Google/Cloud\ SDK/google-cloud-sdk/bin/gsutil.cmd $@
   }
   gcloud() {
   /c/Program\ Files\ \(x86\)/Google/Cloud\ SDK/google-cloud-sdk/bin/gcloud.cmd $@
   }

7. Run the following script to create a new project:

   ./gcp_onboarding.sh -p <PROJECT-ID> -v -r <REGION-CODE>

   where

   <PROJECT-ID>	A unique identifier for your Google Cloud Platform project.Important: Make sure you provide the same projectID value in the GCP-Project-Id setting in the Centralized Tenant Configuration UI. For more information, see Configuration-settings-E-M.
   <REGION-CODE>	A geographical location where you want to execute your Google Cloud Platform machine learning jobs.Important: Make sure you provide the same region code value to the GCP-Region setting in the Centralized Tenant Configuration UI. For more information, see Configuration-settings-E-M.

   New <projectID>-key.json file is generated and saved in the execution folder along with the <privateID>-key.json file.

To use BMC Native (Google) classification for cognitive service

1. Log in to BMC Helix Innovation Studio and navigate to the Administration tab.
2. Select Cognitive and chatbot > Cognitive service.
3. From the Configure list, select Cognitive service connections.
4. In the BMC Native for natural language classification section, in the Service Account Credentials field, add the content from the <projectID>-key.json file downloaded from the Google Cloud Platform.
5. To test the connection, click Test.
   If the connection fails, verify your service account credentials.
6. After verifying the connection is successful, click Save.

Where to go from here

Training-and-testing-the-cognitive-service-for-a-custom-application