Metamaze
Search…
Overview of Uploads
When you open the 'production data' module you will see a list of all uploads with different properties presented in different columns:
    Status - The status of the upload
      Queued (batch upload only) - Waits until all batch files have been sent.
      Ongoing - The upload is being processed
      Manual intervention - The upload or one or more documents are present in the manual intervention module.
      Failed - The upload failed.
      Done - The upload is perfectly processed
    Upload user - Who uploaded the files (user name if uploaded via UI or system if API)?
    Completed date - When was the upload processed?
    Upload id - The identifier of the upload
    Manual intervention - who has carried out manual intervention. Here are the initials of each user who has performed a manual intervention. The initials have a green color if the upload was processed correctly, a yellow color if the manual intervention is still in progress.
    Process steps - These are explained in more detail
    Actions
      Plus icon (start row) - Show all documents with their properties formed from the files in the upload
      Eye icon - View details of the upload
    Checkbox - send the upload to the training set to enrich your model with extra training data from production. Send one, multiple or all (on the same page) at the same time to training. When an upload was sent to training, this will show as an arrow. Sometimes it's not possible to send an upload to training, this is the case when for instance one or more documents are in human validation.
It's possible to customize your view and add/delete columns you would like to see/not like to see, by checking/unchecking a column name in the list which opens when you click on the '+' icon.
Notice there is as well a plus icon you can open at the upload rows, this will give an overview of all documents that are formed from the files of the upload. This sub table has a number of fields:
    Status
      Ongoing
      Manual intervention
      Done
      Failed
    Document type
    Manual intervention
    Document steps - more info below
The table has several filters and search fields:
    Status
    Already sent / not yet sent / cannot send to training dates
    Upload identifier

Process Steps

When an upload is processed, you can follow all steps in the process in real time. Each step is represented by a line, which can appear in three colours
    Orange - The step is in progress
    Green - The step has been successfully completed
    Red - The step has failed.
In an upload row you have two steps:
    Input from upload
    Output of upload
With a document you have four steps:
    OCR
    Document creation (page management and/or document classification)
    Entity extraction
    Business rules
When human validation is in progress or has happened, the line also has a dot:
Example 1
The status upload shows that the upload is currently in human validation. We also see that the input was successful (green line). When opening the upload row, we see more information about the documents. We see that all documents are in human validation (status) and that Niels Sales (NS) has already handled one document (green line with dot), and 4 more have to be done (orange line with dot).
The table style depends on the configuration of the steps of your processing flow. As a second example we show a project that also has page management but no entity extraction:
Because page management is enabled, and therefore each uploaded file, can consist of multiple documents, OCR is performed at upload level instead of at document level. Page management and document classification are also performed at upload level (see the icon with several pages and the number 1). Afterwards, the documents are recognized. Because no entity extraction is activated, there are no steps for each document.
In this example we see that the input and OCR were successful, and that the upload is currently in manual intervention at the document creation step (page management and document classification).

Merge Training Data With Production Data

Via the checkbox, you can send uploads that have been processed to the training data so that your models are enriched with production data and improve your models. You will find the upload in the management module, part uploads, at the top. Through the filter 'source' you can only view training uploads from the 'production pipeline' module.
Last modified 2mo ago