Configure Document Verification

What is Documents?

Documents is a collection and/or verification step that you can configure as part of your company’s user diligence or business diligence processes, such as providing proof of address, proof of SSN, proof of EIN, and more. Each Documents collects a document (SSN cards, W2s, IRS notices, utility bills, etc.) from the end user to extract, process, and verify documents according to your specifications.

Learn more about Documents here.

Verification Template

Just like other verification templates, a Document Verification Template defines the configurations that determine whether documents will pass or fail the verification step.

Unlike other verification templates, Document Verifications have the added value and complexity of being able to extract values or attributes from a large array of documents that may or may not have a standard format or structure. Persona has curated and built models to tackle the vast variability of many of the documents that businesses find useful when operating in a digital space, and these configurations can be set to meet your specific business needs through the Document Verification Template.

The Document Verification Template has two main types of configurations:

• Verification checks
• Extractions

Verifications Checks

Within the Verification checks section at the top of the screen, you will see a list of verification checks that every document runs through. In addition, the output of each check within an inquiry can be reviewed within Dashboard → Inquiries → All Inquiries.

Next to each Verification check is a checkbox labeled Required to pass. If this checkbox is enabled, the Verification Check will have to pass for the document to be accepted (e.g., in the screenshot above, Compromised submission is Required to pass. This means that if the document is a Compromised submission (i.e., the document is found in a publicly available source), Persona will reject this document). Depending on your business needs, you can configure which verification checks should be required.

Extracted properties detection verification check

You may notice that Extracted properties detection has different settings than other Verification Checks. For example, we can configure Extractions below and add them to this check to ensure that given extractions are found before passing the document.

In this example, the extraction named Owner Name 1 is required for the Document checks to pass.

Extractions

Persona can extract and verify specific information from a document; we refer to this specific information as an extracted attribute. More commonly this is used for documents like utility bills, proof of address, proof of income, or non-government-issued identification documents. All extractions are dependent upon the quality of the images provided.

Creating an extraction

1. Navigate to a specific Document Verification Template by either:

   1. Clicking Inquiries > Templates > Selecting the desired Inquiry template (that contains an existing Document Verification Template) > Configure in the top right or
   2. Clicking Verifications > Templates > Selecting the desired Document Verification Template

2. Click on Add extraction

3. For the extraction, decide what whether you’d like to compare the extraction against a known value or not.

   1. When configuring an extraction within a Document step, by default, the extraction will only extract the value. But you can also set it to compare against a known value such as an Inquiry field.
      • Default: The Document Verification will just for the existance of information (e.g. a specific name, any name, a key value pair, regex matched text) within the document
      • Toggle comparison: Toggle an extraction to compare when looking for known user information (e.g., user name, address) in a document or situations in which you already have an existing value that you’d like to compare against. Comparing leverages user-specific information (e.g., individual’s name, EIN) to match and guide the extraction process.
        • For instance, Persona can compare an individual’s name extracted via a Government ID Verification with the same name found in a utility bill Documents extraction in order to verify that the document belongs to the same individual.

   

4. Depending on whether your extraction is extraction-only or a comparison, you must set further configurations specific to the information you plan on extraction. When you’re finished adding those configurations click Add to save your extraction. You can find more details on configurations specific to extraction-only and comparison below.

   

Additional configurations for default extraction method

For the default extraction method (without comparison against a known attribute), you will need to set and choose from the following:

1. Name
   1. Type a label or name that best matches the extraction type you’re looking to extract. This property is used as an identifier and does not affect your extraction in any other way.
2. Require extraction
   1. Check the Require Extraction checkbox to ensure that a specific extraction is matched, before passing the Extracted properties detection verification check. Leaving this unchecked will mean that a specific extraction is not needed to pass the Extracted properties detection verification check
3. Type

   1. Under Type, you can select from four different extractor types.

      

      1. Text: Search for a specific string
         1. Text Pattern: Input a string you wish to search for in a document.
         2. Maximum Edit Distance: Edit distance determines the level of error tolerance permissible in your document. Choosing 0 would require an exact match. It’s recommended that you select a number between 0-2.
         3. Extract Action
            1. Accept: Finding the Text Pattern constitutes a match.
            2. Reject: Matches on the Text Pattern are filtered out from matches that are accepted.
      2. Preset: Detect instances of information units like person, organization, and location names and various numerical expressions like time, money, etc. Persona uses AI and machine learning models to intelligently detect instances of entity types labeled Preset. Entity types are a variety of information units like person, organization, location names, and various numerical expressions like time, money, etc.
         • Preset:
           • Choose from a variety of different entity types. Persona will automatically extract all instances of selected Presets in a given document.
      3. Regex: Use regular expression to search for a specific text pattern
         1. Regex Pattern: Input a valid regex pattern to match against the extracted text from your document.
         2. Extract Action
            1. Accept: Finding the Regex Pattern constitutes a match.
            2. Reject: Matches on the Regex Pattern are filtered out from matches that are accepted.
      4. Key Value Pair: Scan for search keys, which are a list of field names or labels. Persona will then automatically scan for the value below or to the right of the search key if it is found in the document.
         1. Search Keys:
            1. Search Keys are a list of field names or labels ****that Persona will intelligently search for in a given document to identify the corresponding value. For instance, if a document contains the text “Name: John Doe,” listing “name” as a Search Key will allow Persona to intelligently scan the document for the corresponding value (i.e., “John Doe”).
            2. Persona will automatically configure settings to scan for the value below or to the right of the search key if it is found in a document.

Additional configurations for comparison extraction method

For comparison-based extractions, you will need to set and choose from the following:

1. Name

   

   1. Click on the dropdown in order to select a field name. This is an inquiry field that will be used to compare against the extraction (e.g., an individual’s Address City extracted via a Government ID Verification).

2. Require extraction

   1. Check the Require Extraction checkbox to ensure that a specific extraction is matched, before passing the Extracted properties detection verification check. Leaving this unchecked will mean that a specific extraction is not needed to pass the Extracted properties detection verification check

3. Type

   1. Select from a variety of pre-configured types curated by Persona to perform the best possible extraction for the attribute.

4. Minimum match requirement

   1. Under Match Requirement, we have three options:
      1. Full: The match should be exact (e.g., abcd is a full match for abcd). Persona can handle most common abbreviations found in street names (e.g., St.) and titles (e.g., Ms.) when considering an exact match.
      2. Partial: Persona allows for some variations in the text (e.g., abc is a partial match for abcc).
      3. None: Any match (or the lack thereof) will be considered a match. This setting should be used sparingly in most Document checks.