Users often use search keywords that are not recognized by semantic systems leading to no results. By learning from the user search behavior, Actonomy now solves this problem : all new or unknown search terms are automatically added using machine learning. The result : a system that systematically adapts itself to the user context with goal to systematically provide better results not found by traditional semantic search systems.
Powerful and accurate semantic search solutions rely on a thesaurus or ontology to classify and understand what people are searching for. If words are not in this thesaurus and used while searching, then typically semantic search will not give meaningful results or even worse no results at all. “Actonomy’s core ontology has been the largest ontology on the market for years”, says Filip De Geijter, CEO Actonomy. “With the new functionality, we now allow the core system to keep growing and learning from the user.”
Actonomy’s core ontology does not only consist of a standard thesaurus or dictionary in which words are stored. It stores words with the context in which these are used – ‘network architect’ has a totally different meaning than an ‘architect’ as the context is designing ‘networks’ and not ‘houses’. High accurate searching and matching requires millions of words within their context. “We do not only store words but we also model the career cluster information and the context of the words”, says Filip De Geijter, “Now we are learning new words and new contexts providing the user with even better results and avoiding the cost of maintaining the ontology.”
Adaptive semantic search & match is based on advanced combinations of machine learning, neural networks and AI. By setting up the adaptive semantic search as a cloud service, Actonomy clients to not only avoid investing in own platforms but have the benefit that the system learns quickly and adapt very fast to new trends, new search terms and new ways of looking for candidates and vacancies – used by all Actonomy clients and their users!
