As a native content discovery product, Taboola needs to be able to pace the advert request with profile matching Algo through its real time recommendation engine.
Taboola’s Mapping Behaviour of advert request with publisher list is done through Profile Matching recommendation.
With Tabbola Tech getting smarter, knowing whether it is a match made in heaven is only a few seconds away. This algorithms have two data sets with multiple variables that could potentially fit well together.
It may be matching an Advertiser Request for a Publisher on the basis of multiple attributes like Location, Budget & Publisher Monthly visits.
APPROACH
In order to arrive at better rank for a publisher is sent with a HTML code which basically creates an ad space for a fulfilment of advert request, merely matching the publisher profile attributes with advert request and eventually pulling the data extract from the JS-PIXEL.
- Develop spiders to collect data from multiple publisher sites (Manually onboarded with Taboola) on the Internet to quantify the quality of publisher and predict the match between the publisher user profile with prospective scale for earning against Advertiser Campaign (Advert Request)
- Genetic Taboola Algorithms use hybrid models using primary attributes of an advert request with predicted match of publisher profile and its earning model.
BASIC TECHNIQUES, TECHNOLOGIES, TOOLS
- Profile Analysis, Hybrid Models, Genetic Algorithm, Conditional Fields, R, jsoup
- For step-by-step guides to understand the product wise flow:
- Coogle It: https://coggle.it/diagram/X8-7_XFwSnm…
Taboola Profile Matching Recommendation Module-Case Study
Here is a detailed case study on Taboola Product behaviour and profile Matching Structure through a case study.
Leave a Reply