Name Matching Helps Social Media Companies Verify User Identity

Bad Actors Are Using Social Media to Spread False Information

Recently a U.S. presidential candidate called for name verification in all social media profiles. She was mainly concerned with the problem of fake accounts such as Russian or Chinese bot farms or state organizations that spread propaganda and that represent a threat to US national security.

Particularly paired with the rise of AI, bad actors now have the capability to create very convincing narratives about almost anything and disseminate them at scale. We already saw this in Russian attempts to influence the 2016 presidential election and now we are seeing it in pro-Hamas and neo-Nazi posts on social media platforms.

Social Media Companies Face a Trust and Safety Issue

From the standpoint of social media companies, such malign developments are a threat to their brands. They need to keep their platforms trusted and safe. Effective technical means must be developed to minimize the amount of false information in their feeds. Otherwise, user trust will decline, advertising revenue could be impacted, and ultimately the companies’ valuations will drop.

Currently, social media companies are largely protected from liability for any disinformation on their platforms, but that could change if the amount of such material continues to rise and societies see baleful effects. Regardless, fake accounts are a problem for a social media company’s reputation and valuation.

Social Media Companies Are Increasingly Offering Identity Verification as an Optional Feature

Recently LinkedIn rolled out new ways to perform identify verification. Also Meta and X, formerly Twitter, have released verification services. The precise technical and business approaches used vary (Meta and X both offer the coveted blue mark for a fee, while LinkedIn does not charge anything), but the core idea is to make sure that the users posting on the platform are real human beings with a verified identity.

To verify users’ identities, the companies require a Government-issued identity card, place of employment, email, phone number, physical address, birth date, etc. Of course, the precise information required will vary from company to company.

Name Matching Is a Technology That Can Help

A problem arises when a person submits their verification information, and there is a mismatch between the verification information and the information the user registered with. For example, the name submitted on the Government-issued ID may be different from the name manually entered on the platform. Examples of such variation include:

• Nicknames instead of the full name: Robert vs. Bobby vs. Bob
• Different order of name constituents; for example, Asian names usually put the surname first (Park Jae-in vs. Jae-in Park)
• Missing components. For instance, Spanish surnames typically consist of two components, the father’s surname and the mother’s; the latter is often dropped (Santiago Ramos Meseguer versus Santiago Ramos).
• Spelling variations and misspellings
• Abbreviations in items like addresses (Avenue versus Ave., Street vs. St., etc.)
• etc.

The documentation may even be in a script other than Latin. For example, a Facebook user in the Middle East may send in a driver’s license written in Arabic script where the profile name on the platform will be in Latin script.

Name Matching, also sometimes known as Entity Matching, is an AI technology that solves all of these problems.

How Name Matching Works

Name Matching is a technology that matches the submitted data against what’s in the profile on the platform. It matches each individual field such as name and address and provides a score for likelihood of identity. It combines the individual field scores into a unified score. A good name matching tool is highly accurate, trying to minimize both false positives and false negatives.

For each data field, Name Matching needs to apply matching that is suitable to the characteristics of that field. For example, person names need different matching techniques than physical addresses.

Business rules can then determine the appropriate threshold score, and for borderline cases whether to request that the user or resort to a human reviewer.

Name Matching Can Help Make Social Media Become Safer from Bad Actors

There is increasing public pressure to significantly reduce false information on social media and to create safer and more trusted social media platforms. Name Matching is an AI technology that can help the platforms meet this challenge.