Interpreting social data is like
interpreting genetic data
Correlor’s Social Character Recognition (SCR) engine extracts and analyzes a vast array of social signals from users’ social profiles to learn about their personality.
Our childhood environment and education level; the books, music and movies we like; our online social interactions, the types of places we check into — these and other social data points show the human personality behind our social profile.
Such massive data contains powerful hints that predict user behavior. But it is heterogeneous in nature and a huge headache to analyze, with serious limitations such as multimodality, noise, sparseness, etc.
Correlor analyzes this data in a way that gives businesses the results they need. In the same way that biological genes encode both the similarity between different people and the uniqueness of each individual, our Social Genes encode a user's social DNA. Correlor’s SCR engine enables businesses to tailor their approach to users in real time, based on the users’ Social Genes.
Infection
The Social Genes in a user’s social profile define that user’s personality, but these genes aren't static. Like in biology, Social Genes may become “infected” — in other words, distributed and attributed to other objects.
In nature, genes are constantly being transferred between species — humans, animals, virus particles, bacteria, plants, and so on. Social Genes follow the same pattern. They can “infect” web “species” by transferring a social genome onto any web object — a piece of content, a photo, a music file, a location or another person, giving that item or user a characteristic such as “geeky,” “humorous,” “intellectual”, “edgy” or “spontaneous.” It’s like the web object or user inherits these characteristics.
Our SCR engine monitors infection distribution, particularly in cases where different Social Genes have infected the same object, and applies powerful statistical tools to find the affinity between the object and various personality types. The object is assigned its own Social Genes based on the collective infection by the users who have consumed it. This “infection” of an item’s Social Genes does two important things:
- It enables marketers to put a face on a product and target it to customers based on the psychographic traits of the people who like it;
- It makes personalization easier. The task of matching the right product to the right person is done seamlessly, based on that person's personality traits.
Advantages
This approach has three advantages:
- It incorporates personality aspects into the personalization system. Since personality is a critical factor that influences how people make decisions, incorporating personality aspects into the personalization system improves both the quality of the recommendation and the user experience. Unlike statistics-driven prediction, understanding users’ personality traits and applying them to the matching process improves the accuracy and robustness of a personalization system’s predictive ability.
- It provides transparent personalization. Social Genes define psychographic traits, and everyone can understand what they are. Every person, group of people, piece of content or product (that has been “infected” with Social Genes) possesses a social genome that can be shown and explained. Unlike other algorithmic correlations, personalization with Social Genes can be understood easily. The rationale behind the item recommendations is clear and transparent. Also, in certain scenarios the users themselves can tweak their social genome by turning the genes on or off, augmenting the engine’s understanding of their intent and generating recommendations that are even more fine-tuned.
- It tackles some of the current difficulties of personalization.
Today, personalization is far from perfect. In some cases it cannot even be applied — for example, in a “cold start,” which involves new users who have no purchase history, or new items that have no usage patterns. Correlor solves this problem by harnessing data signals from users’ social profiles and offering recommendations based on their Social Genes. In the case of a new item, Correlor uses infection by Social Genes to attribute personalization logic.
Another problem is the personalization of “indecipherable” or uncategorized content — for example, user-generated content such as videos, talkbacks, social interactions and photos as well as visual concepts, humor and dating experiences. Correlor’s SCR extracts the genetic attributes of each uncategorized item and matches them with users’ genomes.
Social Genes:
creation and evolution
The decoding and interpretation of genetic data from large populations is a well-known scientific challenge. Thousands of top-notch bioinformatics experts all over the world face this challenge every day.
But Correlor does not reinvent the wheel. Our approach to decoding and interpreting social data uses recent algorithmic advancements in bioinformatics.
Algorithmic processes in bioinformatics help us recognize the unique patterns that influence a person’s physical characteristics. In biological language, these patterns are called genes.
Just like in biology, we at Correlor work to identify the unique patterns that influence a person’s social characteristics. We call these patterns Social Genes.
Like biological genes, Social Genes have a proven influence on an individual’s characteristics.