Enhanced Human Based Genetic Algorithm
At the core of Xperscore’s API is the E-HBGA, an algorithm validated by a 10,000 participant study. E-HBGA distills all user interactions, or “signals”, into usable intelligence to determine a user’s behaviour, and ultimately, their expertise. At any given time, the expertise network is being refined by new inputs and actions by users, as the algorithm re-balances and redistributes the impact of each signal.
Xperscore’s proprietary Enhanced Human Based Genetic Algorithm features numerous improvements on original HBGA technology, inspired and validated through exhaustive in site testing and experimentation.
R&D Study: 10,000 Students and Academics
In order to generate a pure and reliable set of user interactions, Xperscore led a study involving over 10,000 university students participating in a live Question and Answer environment powered by the E-HBGA algorithm. Hundreds of academics were directly employed to validate results, and correlate the identification of experts with their genuine credentials and experience.
Over the course of the study, as further elements were introduced to the E-HBGA algorithm, accuracy increased from a baseline of less than 60% to over 80% to start. With post-study analysis continuing to suggest additional weighting factors, and given the self-evolutionary nature of our engine, we have strong scientific evidence that led us believe that our accuracy rate improves over time to above 90% in real-world expertise identification.