Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.

Marketers are like anxious lovers. They want to know every little detail about you.

Today, data management platform (DMP) Lotame is offering them more pieces of you. It is announcing a new Semantic Classification Tool that it says offers “greater granularity” about the content of web pages you visit.

David Blumberg, director of product marketing, told me that the new tool employs language analysis, artificial intelligence, and machine learning to scan the page and categorize it in detail.

He said that, instead of just relying on the page’s metadata — like navigational breadcrumbs or search keywords — as some content analyzers do, the tool can assess that you’re interested in, say, “fine dining restaurants on the Upper East Side.” By contrast, the company said, other services may only tell the publisher that you looked at a page about restaurants in New York.

The company says it can detect thousands of unique keywords across pages, which can be used as “targetable behavioral attributes,” so that client companies can create their own customized hierarchies about you, if they wish.

To illustrate, Lotame had its tool scan a page chosen at random. (Because it isn’t a client’s, they asked it not be identified.) The tool generated nearly 20 descriptors: acupuncture, Beirut, Lebanon, traditional Chinese medicine, acupuncture point, energy therapies, religion, wellness, health and medicine, alternative health and medicine, healthy lifestyle, depression, Middle East, Islam (religion), Arabic (language), women’s health, demo: gender: woman, traditional Middle Eastern clothing, Ramadan, and yoga.

Lotame is oriented toward the needs of publishers. You visit a particular page on the publisher’s website, and the tool adds attributes to your anonymous profile for that site, which is stored in a Lotame cookie (a small file) on your computer.

If you had visited the restaurants/Upper East Side page, for instance, the next page you view on that website might show an ad for a high-end French restaurant in the East 70s in Manhattan.

An advertiser that wants to use that info needs the publisher’s approval, Blumberg pointed out, since the data is siloed by the client.

As a cookie-based publisher profile, the info is employed for web and mobile web targeting. Like some other DMPs, Lotame also offers a technology that tries to match the web cookie with a visitor to a mobile app, where cookies aren’t used.

This would allow, for instance, an ad to be targeted to a visitor to CNN.com whether that visitor was on a desktop, a laptop, or the mobile web on a smartphone, as well as in CNN’s mobile app.

This Lotame tech, appropriately called Cross-Device, uses IP addresses, device IDs, operating systems, locations, and other clues to infer probabilistic matches between a cookied profile and the mobile app user. Blumberg said it was 70 to 95 percent accurate, with the accuracy decreasing as the number of users to be matched increases.

Semantic classification of web pages is not uncommon among companies helping to track users by interests. Blumberg told me that Lotame is unique in that the tool is integrated with the company’s other data products, and employing it is “as simple as flipping on a switch.”

This “synergy,” ability to cross-match the user on mobile devices, and classification granularity, he said, make Lotame’s new tool different from similar point solutions offered by Proximate or Sizmek’s Peer39.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.