How to do Auto Extraction and Manual Extraction work?

Data Extraction, also known as Indexing, is the process of extracting data from a document. Indexing enables data extraction, digitization, validation, amending document status, GL coding, and data storage.

There are two types:

1. Auto Extraction
2. Manual Extraction

Auto Extraction

Auto extraction is when Capture Point automatically identifies data such as invoice number, invoice date, vendor ID, etc., using OCR and Machine Learning technologies. This happens when a document is opened on the Document Indexing page. If the expected fields are not identified automatically, the user performs manual extraction that trains the application to extract data of the same format in future documents.

1. Click  the UpArrow icon to view the document indexing page.
2. Capture Point auto-detects the data such as invoice number, and invoice date, and reflects them in respective fields on the left pane.

Confidence Indication:

On the indexing page, each field indicates a color code for the confidence level of extracted data.

• Green for high confidence 90% -100%
• Yellow for medium confidence 80% – 90%
• Red for low confidence below 80%

Manual Extraction

Manual extraction is performed only if the expected fields are not extracted automatically or if there is an error in the extracted data during auto extraction.

To perform manual extraction,

1. Click the respective field from the left panel. The field is now highlighted with a blue border.
2. On the right panel, locate the document, and surf through the document. Red rectangles are shown hovering over the OCR-enabled fields.
3. Click the value to populate inside the field in the left panel. In case of multiple words are included, the user should drag a box around the words.
4. Capture Point learns from the manual extraction and auto extracts the fields in the future.