How does your semantic search work?

Our semantic search engine is fine tuned to understand the context and focus of your search input. Your search input, e.g. text or number is reduced to a standardized fingerprint representation which we can then use to compare your input against our database of more than 120 million patents. We can then calculate a similarity score between your entered fingerprint and the fingerprint of any patent family. That way, we can easily prioritize your results by similarity. The comparison of your input to the results, as well as the input itself are the two major differences to a classic Boolean search: the Boolean method is a binary operation that checks whether something is true or false -so whether the rule your search query represents is met or not. A similarity-based approach is a comparison on a continuous spectrum: there will likely always be some similarity between two texts, as long as there are some words that are similar. As our search engines extract much more information from your input than you would likely have in a Boolean input, it is a more fine-grained comparison. And as we cut off the results after the top 1000 most similar ones, you have a pre-selection that you can then filter down on using our interactive graphs as simple to use Boolean filters.