DeepOpinion Document Information Extraction

    • Ahmed Al-Ali |
    • 0 steps |
    • 5 minutes
    Navigate to your workspace on [](
    Navigate to the Cognitive Skills Builder using the hamburger menu on the top left
    Click "BUILD A NEW SKILL" to create a document information extraction model Select "Document Information Extraction" Click this text field and type a name "Invoice Information Extraction" Click "CREATE"
    This will take you into the project to start. Now you can upload your documents. Upload 1 document for now.
    Once the upload is done you can move to the next step
    Navigate to "Label Definition" from the side menu to list the information you would like to label and extract from a document
    A label will be automatically created with the name "default" You can edit it and rename it to "Invoice#" You can also change the label color
    You can also choose to upload the list of labels as csv and change the color of each label for easier differentiation
    Navigate to "Training Data" to start data labelling
    Now you need to select the uploaded document for labelling. This step will automatically take you to the document labeling page
    First step, make sure to select the curser for ease of navigation. Other tools are used for more advanced labelling,
    IMPORTANT Once done labelling, click "Update" to save your annotations
    To label text on the document, first click on the bounding box then click on its corresponding label on top of the document. This will assign a label to it automatically
    Now do the same for all the labels that are available on this document. For text that is split into multiple bounding boxes, just highlight all of the and then select the corresponding label.
    You can check for your labels by looking at the right hand side widget and compare the labels that were assigned to the text
    COMING SOON! Once you finish labelling enough documents you can navigate to "Cognitive Skill Training" from the left menu to train a model.

    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