Customer inquiry analysis for online customer service scenarios

Activate the service and purchase a resource plan

Before you start, ensure that the service is activated. After activation, you can purchase a resource plan.

• Activate the service: Activation page

• Purchase a resource plan: Purchase page

Service invocation and testing

For more information about model invocation, see Model invocation.

For software development kit (SDK) examples, see SDK examples.

Debugging

Configure access credentials using environment variables

1. Notes:

   1. An AccessKey for an Alibaba Cloud account has permissions for all APIs, which poses a high security risk. We recommend that you create and use a Resource Access Management (RAM) user for API access or daily operations and maintenance (O&M). You can log on to the RAM console to create a RAM user.

   2. Do not save your AccessKey ID and AccessKey secret in your code. This can lead to credential leaks. Instead, configure environment variables to store and access your credentials.

2. Configure for Linux and macOS

   export NLP_AK_ENV=<access_key_id>
   export NLP_SK_ENV=<access_key_secret>

   Replace <access_key_id> with your AccessKey ID and <access_key_secret> with your AccessKey secret. For more information about how to obtain an AccessKey, see Step 2: Obtain an AccessKey for your account.

3. Configure for Windows

   1. Create a new environment variable file. Add the NLP_AK_ENV and NLP_SK_ENV environment variables. Then, set their values to your AccessKey ID and AccessKey secret.

   2. Restart your Windows system.

Java code example

/**
 * An AccessKey for an Alibaba Cloud account has permissions for all APIs. This is a high security risk.
 * Create and use a RAM user for API access or daily O&M. Log on to the RAM console to create a RAM user.
 * This example shows how to store the AccessKey ID and AccessKey secret in environment variables.
 * You can also store them in a configuration file as needed.
 * Do not hardcode the AccessKey ID and AccessKey secret in your code. This can lead to credential leaks.
 */
String accessKeyId = System.getenv("NLP_AK_ENV");
String accessKeySecret = System.getenv("NLP_SK_ENV");
DefaultProfile defaultProfile = DefaultProfile.getProfile("cn-hangzhou",accessKeyId,accessKeySecret);
IAcsClient client = new DefaultAcsClient(defaultProfile);
Map<String, Object> map = new HashMap<>();
map.put("input", "Your service attitude is poor. Is this how you provide service?");
RunPreTrainServiceRequest request = new RunPreTrainServiceRequest();
request.setServiceName("Dialog-Analysis");
request.setPredictContent(JSON.toJSONString(map));
RunPreTrainServiceResponse response = client.getAcsResponse(request);
System.out.println(response.getPredictResult());

Python code example

# Install dependencies
pip install aliyun-python-sdk-core
pip install aliyun-python-sdk-nlp-automl

# -*- coding: utf8 -*-
import json
import os

from aliyunsdkcore.client import AcsClient
from aliyunsdkcore.acs_exception.exceptions import ClientException
from aliyunsdkcore.acs_exception.exceptions import ServerException
from aliyunsdknlp_automl.request.v20191111 import RunPreTrainServiceRequest

# An AccessKey for an Alibaba Cloud account has permissions for all APIs. This is a high security risk.
# Create and use a RAM user for API access or daily O&M. Log on to the RAM console to create a RAM user.
# This example shows how to store the AccessKey ID and AccessKey secret in environment variables.
# You can also store them in a configuration file as needed.
# Do not hardcode the AccessKey ID and AccessKey secret in your code. This can lead to credential leaks.
access_key_id = os.environ['NLP_AK_ENV']
access_key_secret = os.environ['NLP_SK_ENV']

# Initialize AcsClient instance
client = AcsClient(
  access_key_id,
  access_key_secret,
  "cn-hangzhou"
);
# The input can also include preceding (context_above) and succeeding (context_below) context information to improve the algorithm's performance. This information is optional. See the input example.
content = {
  "input": "Your service attitude is poor. Is this how you provide service?",
}
# Initialize a request and set parameters
request = RunPreTrainServiceRequest.RunPreTrainServiceRequest()
request.set_ServiceName('Dialog-Analysis')
request.set_PredictContent(json.dumps(content))
# Print response
response = client.do_action_with_exception(request)
resp_obj = json.loads(response)
predict_result = json.loads(resp_obj['PredictResult'])
print(predict_result['result'])

PredictContent example

# This is a complete example that includes preceding (context_above) and succeeding (context_below) context data. This data helps improve the algorithm's performance, but it is optional.
{
  "context_above": [
    {
      "role": "User",
      "context": "I asked you to change the price and you just canceled my order?"
    },
    {
      "role": "Agent",
      "context": "Orders cannot be modified after they are placed."
    }
  ],
  "input": "Your service attitude is poor. Is this how you provide service?",
  "context_below": [
    {
      "role": "Agent",
      "context": "I'm sorry, but that is the policy."
    },
    {
      "role": "User",
      "context": "Goodbye. I'm never coming back."
    }
  ]
}

PredictResult example

{
  "emotion": {
    "key": "Complaint",
    "score": 0.4929790496826172
  },
  "intent": {
    "key": "None",
    "score": 0.756518542766571
  },
  "category": {
    "key": "Other-Other",
    "score": 0.4580000042915344
  },
  "sentiment": {
    "key": "Negative",
    "score": 1.0
  },
  "aspectItem": [
    {
      "aspectCategory": "Agent-Service",
      "aspectPolarity": "Negative",
      "negativeProb": 1.0,
      "positiveProb": 0.0,
      "terms": [
        {
          "aspectTerm": "Service attitude",
          "opinionTerm": "Poor"
        }
      ]
    }
  ]
}

Input parameters

Response parameters