The following article will explain Model Details' customisation in KAPTO. To eliminate confusion, the article will include a screenshot with annotated descriptions to show what the Model Details in KAPTO look like.

1. Navigate to Models.

2. Open the model.

3. Click the Edit model button.

The following section will provide more information about:

‍Model Name: The name given to a KAPTO model with regard to uniquely identify from other models; may contain a naming convention based on elements that are important to the model designation.

Tags:  Assign tags to describe a KAPTO model to be retrieved by searching.

Manager: Assigned User Role to a KAPTO user.

State: It is important to select the appropriate state of the model, which can be either "In training" or "Operational." The state of the model reflects the current stage of development and the intended use. If the model is still being trained and refined, then it should be set to "In training."

During this stage, the model is still learning from the data it's being fed and improving its accuracy. Once the model has been thoroughly trained and is ready to be used for its intended purpose, it can be set to "Operational." At this point, the model is fully functional and can accurately classify new data it's presented with.

Model Type: This entitles the Classification or Extraction Model in KAPTO. To cover additional information that complements the text of documentation.

Classification Model Type: The KAPTO Classification Model concludes from the input values given for training. The model identifies the categories into which new data will fall into various classes. Classification can be performed on structured or unstructured data.

Extraction Model Type:  The objective of the extraction model is to read documents, extract information, and populate entity instances with the extracted values.

In the following sections, we will delve into the concepts of Summarization, Table Detection, and Form Detection.

Summarization: The goal of summarization is to condense large amounts of text while maintaining its important content. Some models can summarize text by extracting key information from the original text, while others can generate entirely new text based on the input.

Table Detection: The extraction phase evidently is implicated in a traditional OCR recognition process. However, the entities are structured in rows and columns instead.

Form Detection: Used to extract text, key-value pairs, tables, and structures from documents automatically and accurately.  

Template: We can define templates, and through categorization, AI finds that these templates automatically belong to this category. This is one option for automatically using the information; the second is that when the document category that belongs to this type of page layout is attached, the pre-processing phase is making image registration. In the case where a template is used, it means that the image is registered over the template.

‍