Customer Support Categorization -Deepopinion | Scribe

    Customer Support Categorization -Deepopinion

    • Ahmed Al-Ali |
    • 0 step |
    • 6 minutes

    1. Create A Label Model

    Log-in to [https://studio.deepopinion.ai/](https://studio.deepopinion.ai/) Navigate to your collaboration workspace on the top left click on "My Workspace" and choose the assigned workspace for your organization to collaborate.
    From the Dashboard page click on "CUSTOM MODELS" create a new label model
    DeepOpinion Studio supports multiple types of text analysis models. In the case of customer support automation there are multiple tasks to accomplish one of which is assigning a ticket to the responsible team "ticket routing". For such use case a "Label" model is the suitable option. Label Model : "Assign exactly one label to a piece of text, e.g. routing support tickets to teams"
    On the Custom Models page you you can create a new AI model Click "CREATE A CUSTOM MODEL" on the top right Give your model a name "Customer Support Categorization (Banking)" As we want to assign a single choice to customer tickets select "LABEL" from the dropdown menu Click "CREATE"

    2. Upload Training Data

    Tip! Historical data from customer support tickets can be used to train an AI model to perform the same tasks. This way the AI model will learn to assign the right ticket priority, type, team, tags, and sentiment based on what employees did in the past to similar tickets. Note: for this tutorial we provide you with a sample of publicly available dataset from a banking customer support tickets.
    Upload training data for the AI model Navigate to Step 3 and click "Training Data" Click "UPLOAD" select the file "Customer_Support_Categorization.csv" Select the first column "text" as your customer support ticket Select the second column "category" as the ticket category Click "CONFIRM AND UPLOAD TRAINING DATA" Now your training data is being uploaded and prepared for training the AI model.
    You can have a look at the uploaded training data by navigating back to Step 2 "Label Definition". There are 77 categories listed with example under each label dropdown menu. There is no action required at this step.

    3. AI Model Training

    Tip! DeepOpinion Studio offers various specialized industry specific and language specific proprietary models. Selecting a suitable base model contributes to enhancing the accuracy as it is already trained on similar domain and context (Transfer Learning). Some examples are banking industry, consumer goods industry in English, German, and Arabic languages. If there isn't an exact fit for your use case, the "English Base" is a great default option.
    Navigate to Step 4 "Model Training" to start the AI model training on the data you just uploaded Click "Train a model" Give it a name such as "Banking Support v1" Click on the Base model to train dropdown menu and choose "Banking (id: 86)" Click "START TRAINING" Now your AI model is being trained on the uploaded data.
    Once your model finishes training it will achieve around 90% accuracy score "0.90". You can expect the accuracy score to have a minimal variation. By clicking on it you can explore the model performance and metrics in more details.
    After confirming that you are happy with the model performance, it is ready to be published and used in production. Click on the upward facing arrow to publish your model.

    3. Analyze Customer Support Tickets

    Tip! Once you have a published model you can use it in multiple ways: 1. Analyze a csv or excel file containing customer support tickets by directly uploading it to DeepOpinion Studio 2. Connecting your model to an Automation on DeepOpinion Studio to automate tickets on any of the supported integrations such as Freshdesk and Zendesk 3. Using DeepOpinion API to call this model on any other application For the purpose of this Tutorial we will demonstrate option 1.
    Click "Batch" from the sidebar menu on the left to start analyzing tickets by directly uploading the csv file on DeepOpinion Studio.
    Here you will upload the same file that you trained the AI model on to experience batch analysis of customer tickets Click "Create a new batch analysis" Click "BROWSE FILES" Select the file "Customer_Support_Categorization.csv" Select the first column "text" to be analyzed as it holds the customer support ticket messages Under "Advanced Settings" untick "Split the text in each row into smaller chunks, e.g. sentences" as you want to assign one label to the whole ticket Click "CREATE BATCH ANALYSIS" Now the file is being analyzed where categories are assigned to each ticket.
    You can explore the customer support ticket categorization results on a summary dashboard by clicking on the analysis once it is completed.
    You can also download the results of the batch analysis as an excel or csv file

    Want to make guides like this in seconds? Yes, it's really that fast.

    This Scribe is in tip-top shape!Leave feedback if there are any issues with this Scribe