Skip to Content
Product Information

Get Insights on Historical Tickets with Service Ticket Intelligence

Service Ticketing Systems are responsible for handling large ticket volumes to ensure a smooth and uninterrupted service delivery towards achieving business success. These tickets generally contain valuable information that one has to manually extract, in order to gain insights for further improvements.

The process is often labour intensive and costly but with the clustering (Beta) feature in Service Ticket Intelligence that we released in 2102, meaningful ticket clusters containing concise keywords and information can be automatically generated and retrieved with unstructured textual data. This will enable you to discover insights on hidden information, patterns, trends, hotspots, etc. and use it for improving the reliability, processes, customer satisfaction, etc. of your business.

The options to make use of the clustering feature are in fact limitless, but let’s take a look at the common usage scenarios and find out the value that the clustering feature can bring about below.

Ticket Clustering Overview

Ticket Clustering Overall Process

  1. Data Upload
    • Historical data (tickets) is uploaded to Service Ticket Intelligence
    • The data is required for training a clustering machine learning model
    • A new model will be created and it will be in NEW state
  2. Model Training
    • Once data is uploaded, initiate the model training process
    • Model state will transition from NEW->PENDING_TRAINING->IN_TRAINING->READY
    • This process might take several minutes depending on your dataset size
  3. Model Activation
    • Once model is trained and in READY state, model activation is required before retrieval of cluster information is possible
    • The model would be in ACTIVE state as soon as the model is activated
  4. Inference
    • Retrieve clusters generated
    • Options are available for filtering clusters
      • top k clusters
      • group-by
    • Uploading of new data for generation of new clusters is also possible at this stage, if the context of data is similar
  5. Analysis and Decision
    • Process the cluster information, retrieve insights with keywords, sample tickets, ticket characteristics, etc. and make follow-up actions and decisions accordingly

Examples

1. Data Upload (POST /sti/training/model)

Mapping input determines which data is to used for the generation of clusters

Mapping output determines which data is to be present in sample tickets as well as to be used for filtering clusters

Sample Request Body

{
    "scenario":
  {
      "desc":"Training Clustering travel data (small)",
      "type":"clustering",
      "language":"en",
      "business_object":"ticket"
  },
  "mapping":
  {
        "input": ["description"],
        "output": ["id", "description", "category"]
  },
  "training":
  {
        "file": "<<base64 encoded csv file>>"
  }
}

2. Model Training (POST /sti/training/model/train)

Sample Request Body

{   
    "model_id": "<<model_id>>"
}

 

3. Model Activation (PUT /sti/training/model/activate)

Sample Request Body

{   
    "model_id": "<<model_id>>"
}

4. Inference (POST /sti/text/cluster)

To retrieve all clusters generated from the training data

Sample Request Body

{   }

To retrieve top k clusters from the training data

Sample Request Body

{
   "options":{
      "top_k_clusters":2
   }
}

To retrieve clusters group-by field from the training data

Sample Request Body

{
   "options":{
      "cluster_groupby":{
         "column":"category",
         "value":[
            "compliment"
         ]
      }
   }
}

To generate and retrieve new clusters from the data provided in request body

Sample Request Body

{
  "scenario":
  {
      "type":"clustering",
      "language":"en",
      "business_object":"ticket"
  },
  "mapping":
  {
        "input": ["description"],
        "output": ["id", "description", "category"]
  },
  "clustering":
  {
        "file": "<<new base64 encoded csv file>>"
  }
}

Additional Resources

Postman collection: https://github.com/SAP-samples/service-ticket-intelligence-postman-collection/tree/clustering

API specs for clustering training: https://help.sap.com/viewer/5088c3bb02144e7782959bb1529ca70e/SHIP/en-US/d663408e81b6406f98bb2b14b45f3e6c.html

Examples: https://help.sap.com/viewer/5088c3bb02144e7782959bb1529ca70e/SHIP/en-US/d2b81fc4b8424a0eaa9d6cf420751fe9.html

API specs for clustering inference
https://help.sap.com/viewer/5088c3bb02144e7782959bb1529ca70e/SHIP/en-US/5d7142a96c0f49428323fa75a0a40a3a.html

Examples
https://help.sap.com/viewer/5088c3bb02144e7782959bb1529ca70e/SHIP/en-US/61fdc051905149e398c23d33a9cb8824.html

Beta Feature and Usage

As clustering is released as a Beta feature in Service Ticket Intelligence, you will first need to register your interest at SAP’s Customer Engagement Initiative Project (on clustering) in order to gain access for trying out this feature.

To try out other features (classification, recommendation scenarios) of Service Ticket Intelligence, we are also available on SAP Cloud Platform as a free trial.

Find out how you can set up your own trial account with Service Ticket Intelligence here and give the service a try today!

For more information on SAP AI Business Services:  

Explore: SAP Community Page 

Dive deeper: Open SAP Course  

Get an overview: Blogpost part I  Blogpost part II  

Exchange Knowledge: 

Document Classification Questions Document Information Extraction Questions  

Business Entity Recognition Questions Service Ticket Intelligence Questions  

Data Attribute Recommendation Questions | Invoice Object Recommendation Questions 

Be the first to leave a comment
You must be Logged on to comment or reply to a post.