Over a million developers have joined DZone.

Salesforce and IBM Watson Tone Analyzer Integration Using Mule

DZone's Guide to

Salesforce and IBM Watson Tone Analyzer Integration Using Mule

A Salesforce user can understand the tone of a customer's sent mail by integrating Salesforce and IBM Watson Tone Analyzer service using Mule Soft ESB. Here's how.

· Integration Zone
Free Resource

Modernize your application architectures with microservices and APIs with best practices from this free virtual summit series. Brought to you in partnership with CA Technologies.

In this competitive world, it is very important to understand the linguistic tone of a customer so that we can decide whether we need to respond to the customer immediately.

These days, customer satisfaction is more important to a sales manager or representative. A sales manager or representative may want to understand the linguistic tone from his customer's sent mail. For example, when a customer sends a mail to a sales manager or representative, the sales manager or representative may want to understand the tone of the mail so that the sales manager or representative can decide on further actions. This powerful feature can be achieved by leveraging IBM Watson Tone Analyzer service.

Here we want to show how a Salesforce user (sales manager or representative) can understand the tone of his customer's sent mail by integrating Salesforce and IBM Watson Tone Analyzer service using Mule Soft ESB.

We have written a Mule flow that makes the integration of Salesforce and IBM Watson Tone Analyzer service very easy.


However, below are the prerequisites to test the flow shown below:

  • An account at IBM Blumix.
  • A deployed Tone Analyzer service to IBM Blumix.
  • An account at Salesforce.
  • A custom contact field CutomerTone__c on Salesforce. See here to create a custom field on any Salesforce object.

Flow Steps

  1. Receive mail from the configured mailbox using Mule IMAP connector. Once mail is received, the current Mule Message contains email body as payload and the customer mail address as an inbound header fromAddress.
  2. Enrich the current Mule Message with the Salesforce contact ID corresponding to the fromAddress retrieved above. The enriched Mule Message now contains a flow variable contactId that holds the Salesforce contact ID corresponding to the customer mail address.
  3. Post the mail body held in the current payload to the IBM Tone Analyzer service. The Tone Analyzer service gives a JSON response representing the tone of the mail in the case of success. See here for a sample response.
  4. Calculate the tone based on the rule described here. As of now, we are interested only in Emotion Tone. The Emotion Tone has three children tones named Cheerfulness, Negative, and Anger. Each of these tones has an attribute called normalized_score representing the value of the tone. If the sum of normalized_scores of Negative and Anger tones is greater than Cheerfulness tone, then we consider the tone of the customer to be Angry otherwise Normal.
  5. Update the Salesforce Contact corresponding to the contact ID held in the flow variable contactId with the tone calculated above.

Tone-Analyzer Flow

Below is the main flow that receives an incoming mail from a customer.

<flow name="tone-analyzer">
    <imaps:inbound-endpoint host="${gmail.host}" port="${gmail.port}" user="${gmail.user}" password="${gmail.password}" connector-ref="IMAP_for_Gmail" responseTimeout="10000" doc:name="imap"/>
    <logger message="From Address:  #[message.inboundProperties.fromAddress] - To Address: #[message.inboundProperties.toAddresses] - Subject: #[message.inboundProperties.subject] - Payload: #[payload]" level="INFO" doc:name="logger"/>
    <enricher target="#[flowVars.contactId]" doc:name="enrich-with-contact-id">
        <flow-ref name="get-sf-contact-id" doc:name="get-sf-contact-id"/>
    <logger message="Contact Id: #[flowVars.contactId]" level="INFO" doc:name="logger"/>
    <flow-ref name="get-tone" doc:name="get-tone"/>
    <flow-ref name="update-sf-contact-with-tone" doc:name="update-sf-contact-with-tone"/>

The corresponding IMAPS connector configuration is below.

<imaps:connector name="IMAP_for_Gmail" validateConnections="true" deleteReadMessages="false" doc:name="IMAP" checkFrequency="1000">
    <imaps:tls-client path="*" storePassword="*" />
    <imaps:tls-trust-store path="*" storePassword="*" />

The tone analyzer flow has three sub-flows: get-sf-contact-id, get-tone, and update-sf-contact-with-tone. Each flow is detailed in its own section.

Get-Sf-Contact-Id Flow

This sub-flow retrieves the Salesforce contact ID corresponding to the customer mail address.

<sub-flow name="get-sf-contact-id">
  <sfdc:query config-ref="Salesforce__Basic_authentication" query="dsql:SELECT Id FROM Contact WHERE email = #["'" + org.apache.commons.lang3.StringUtils.substringBetween(message.inboundProperties.fromAddress, "<", ">") + "'"]" doc:name="get-sf-contact-id"/>
  <component class="com.capiot.sf.ContactIdRetriever" doc:name="retrieve-contact-id"/>

To retrieve the contact id, we use Mule Salesforce connector and below is the connector configuration

<sfdc:config name="Salesforce__Basic_authentication" username="${sf.username}" password="${sf.password}" securityToken="${sf.securityToken}" doc:name="Salesforce: Basic authentication"/>

The sfdc:query returns an instance of java.util.Iterator. So we have written a custom java component to retrieve the id.

import java.util.Iterator;
import java.util.Map;
import org.mule.api.MuleEventContext;
import org.mule.api.lifecycle.Callable;

public class ContactIdRetriever implements Callable {
    public Object onCall(MuleEventContext eventContext) throws Exception {
        Map<?, ?> contactIdMap = null;
        if (eventContext.getMessage().getPayload() instanceof Iterator) {
            Iterator<?> iterator = (Iterator<?>) eventContext.getMessage().getPayload();
            if (iterator.hasNext()) {
                contactIdMap = (Map<?, ?>) iterator.next();
            return contactIdMap.get("Id");
        return null;

This sub-flow finally returns a contact ID corresponding to the customer mail address. In the tone analyzer flow, we use an enricher to enrich the message with the contact ID retrieved from Salesforce and this contact id is stored in a flow variable #[flowVars.contactId]

Get-Tone Flow

This sub-flow gets the tone of the mail from the customer. It posts the mail body received from the customer to IBM Watson Tone Analyzer and retrieves the linguistic tone of the mail.

<sub-flow name="get-tone">
  <dw:transform-message doc:name="Transform Message">
    <dw:set-payload resource="classpath:EmailBodyToTonePayload.dwl"></dw:set-payload>
  <http:request config-ref="HTTP_Request_Configuration" path="/v1/tone" method="POST" port="443" doc:name="get-tone">
      <http:header headerName="Content-Type" value="application/json"/>
      <http:header headerName="Authorization" value="${authorization}"/>
  <object-to-string-transformer doc:name="object-to-string"/>
  <logger message="#[payload]" level="INFO" doc:name="logger"/>

The corresponding HTTP Request Connector is shown below:

<http:request-config name="HTTP_Request_Configuration" protocol="HTTPS" host="gateway.watsonplatform.net" port="443" basePath="tone-analyzer-experimental/api" doc:name="Tone_Analyzer_HTTP_Request_Configuration"/>

The corresponding Mule Dataweave transformation is shown below:

%dw 1.0
%output application/json
    "scorecard" : "email",
    "text" : payload

This flow returns tone response in the JSON format.

Update-Sf-Contact-With-Tone Flow

This sub-flow updates the tone of the customer mail to the custome field CustomerTone__c of Salesforce contact ID.

<sub-flow name="update-sf-contact-with-tone">
        <json:json-to-object-transformer returnClass="java.lang.Object" doc:name="json-to-object"/>
        <expression-component doc:name="calculate-tone">
         cheerfulnessTone = payload.children[0].children[0].normalized_score;
         negativeTone = payload.children[0].children[1].normalized_score;
         angerTone = payload.children[0].children[2].normalized_score;
         angryTone = negativeTone + angerTone;
         if(angryTone > cheerfulnessTone) {
            payload = '{"toneScore" :' + angryTone + ', "tone" : "Angry"}';
         } else {
            payload = '{"toneScore" :' + cheerfulnessTone + ', "tone" : "Normal"}';
        <logger message="Tone: #[payload]" level="INFO" doc:name="logger"/>
        <dw:transform-message doc:name="Transform Message">
            <dw:input-payload doc:sample="unknown.dwl"/>
            <dw:set-payload resource="classpath:ToneToContactUpdate.dwl"/>
        <sfdc:update config-ref="Salesforce__Basic_authentication" type="Contact" doc:name="save-mail-tone">
            <sfdc:objects ref="#[payload]"/>
        <logger message="#[payload]" level="INFO" doc:name="logger"/>

The above flow applies some logic on the tone response retrieved from the get-tone flow. This logic is encapsulated in the Mule expression-component. This logic finally gives a JSON payload, represents the tone either Angry or Normal, and is updated to Salesforce contact's custom field CustomerTone__c.

The corresponding Dataweave transformation that transforms tone payload to such a payload corresponds to Salesforce Contact update is shown below:

%dw 1.0
%input payload application/json
%output application/java

    Id: flowVars.contactId,
    CustomerTone__c: payload.tone

So finally, this flow updates tone, either Angry or Normal, to the Salesforce contact.

The Integration Zone is proudly sponsored by CA Technologies. Learn from expert microservices and API presentations at the Modernizing Application Architectures Virtual Summit Series.

mule esb ,integration ,ibm watson ,tone analyzer

Opinions expressed by DZone contributors are their own.


Dev Resources & Solutions Straight to Your Inbox

Thanks for subscribing!

Awesome! Check your inbox to verify your email so you can start receiving the latest in tech news and resources.


{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}