Skip to content

Questions?  ·

How Primal’s artificial intelligence creates meaningful interests data

The most frequently asked question about Primal is, Where do you get the interests data? It’s a fair question. If Primal is the most comprehensive source of open interests data in the world, it begs the question of where we get this valuable data. (1)

Unlike our competitors, we don’t ingest and analyze vast amounts of historical consumer data to derive interests data. Instead, Primal creates this data on a just-in-time basis, with each and every request.

The reason Primal seems to have data about every conceivable topic of interest is that Primal is literally creating the data on-the-fly.

Demonstration Walkthrough

Let’s say you have an interest in recruiting a new member to your team of computational genetics researchers. A quick search over LinkedIn suggests there are very few profiles that match that term directly. But there are over 48,000 that have something to do with aspects of this interest (such as computational biology or genetics). Imagine having to manually sift through thousands of potential candidates across such a complex domain of knowledge!

Here’s how Primal makes this sort of research task dead simple. The following automated process works in real-time. I’m going to break down the process, step-by-step.

Step 1: Input Keywords of Interest

Primal is able to create semantic data for very specific topics. When you provide the keyword, computational genetics researcher, Primal first represents your interest using terms it recognizes (we call these recognized terms, Primal’s vocabulary). A subset of this understanding is visualized in the figure below. (2)

Note that Primal initially knows nothing of your specific topic of interest (yet). It isn’t retrieving the data from some semantic knowledge-base in the sky. Instead, it’s going to create that data from scratch.

Step 2: Create Interest Graph

Using the initial semantic representation of your interest from Step 1, Primal begins synthesizing an understanding of the more complex topic of interest. The rules Primal applies in this synthetic process are simulating the activities that human knowledge engineers use when they create formal knowledge representations. In the visualization below, one of these rules is unpacked to show how Primal composed the narrower topic of a medical computational genetics researcher.

Note again that Primal isn’t retrieving the data from an existing semantic knowledge-base. It’s using its finite vocabulary of semantic data and this computational intelligence to create the data that represents your unique topic of interest. (You can see a snapshot of this creative process in the visualization above: Primal created the intermediate topic of medical researcher based on evidence in its vocabulary about researcher, scientist, and medical scientist.)

Primal repeats this process of meaning-making recursively until it has created a large network of topics related to your interest. This interest graph is one-of-a-kind, created specifically for you, about your unique interest.

In the visualization above, the green dots correspond to terms that were discovered in the target content; the red dots are terms that Primal believes are relevant to your interest, but were not found in the target content. Primal uses these intersections in the target content to validate its semantic data and tailor your interest graph to the content you care about.

Step 3: Filter Content of Interest

Your interest graph is now available to work on your behalf, filtering and organizing content based on your specific interest in computational genetics researcher. To complete the job, we use your interest graph as a filter to organize the unstructured content that is the target of your interest (in this case, LinkedIn profiles).

The figure above illustrates how Primal works with external content on the Web. Your interest graph is the red tree-like graph; for this example, LinkedIn is the information service. Primal retrieves the unstructured content and filters it against the topics in your interest graph, returning you the content ranked and annotated (tagged) with topics related to your interest.

A few of the top results are displayed below. Note that many of the most meaningful terms in the result are different from the original topic of interest, computational genetics researcher. Primal uses the evidence of these semantic terms from your interest graph to score and rank the unstructured content.

Interests Data, Made-to-Order

Primal doesn’t store and retrieve its interests data, it creates it from a much smaller vocabulary of semantic data and a computational intelligence that simulates knowledge engineering. Just as humans can express themselves in limitless ways, Primal can create interests data for a limitless range of knowledge domains and applications.

Getting Started

If you’d like to incorporate Primal’s interests data into your own solution, check out our developers site to get started!


  1. Other companies like Google and Facebook collect vast amounts of data about your individual interests. However, they keep it closed and unavailable for third-party use. Primal’s interests data is open and transparent, accessible to consumers and customers alike.
  2. For this illustration, Primal’s data has been narrowed considerably to make the visualizations easier to understand. Whereas a human’s vocabulary might comprise around 35,000 words, Primal’s vocabulary of semantic terms numbers in the millions. Using this semantic vocabulary within its computational intelligence, Primal can generate high quality interest graphs for virtually any topic.