By: Nasreen Bakht Brady, Client Engagement Director and Devin Salmon, Patent Analyst, IP.com
When do you know when you’re done with a patent search? The cheeky response is ‘well how much time do you have?’ Striking a balance between efficiency and quality is no easy task. Many patent search methods start with all the worldwide patents from all time and whittle down that enormous list with filters for keywords, CPCs, assignees, and so on, until the stack is both relevant and reviewable. The number of results is so important because you can’t possibly review a million patents, and the entire result set must be reviewed because no results are raising their hands as more important than others.
But while whittling down the result set to a manageable number, are you discarding important prior art? Further, is whittling down the right method to really find what you’re looking for?
The case is very different in semantic searching because the results are returned on a relevance scale – in other words, the most important ones are at the front of the line, raising their hands. Rather than whittling down all the patents of the world, the semantic engine understands concepts from the query and finds connections to results with shared concepts – the more connections found, the higher the relevance.
We expect to see many results, even millions, returned in semantic searching because many documents may be related in some way, but we use the relevancy score to guide us. For example, in InnovationQ we use a star ranking system to provide insight into relevancy. Four and five star results are most important and match most or all the concepts from your query. If you are assessing the novelty of a new invention, four and five star results should be reviewed for potential knockout art. Three star results, i.e. those results farther down in the list, might only include some of the concepts or features from the query and might be important for §103 non-obviousness or if you are seeking tangential technology.
It’s dangerous for a search tool to impose a default relevancy cutoff for fear that the cutoff is too high and will eliminate important prior art. It’s a best practice to use a relevancy cutoff dependent on the goals of the search and after preliminary review of the result set.
InnovationQ’s powerful semantic engine understands your query and the concepts present in the documents and matches them, independent of literal keywords used or classifications. For those whittlers out there struggling to embrace semantic searching, it does take a bit of a shift in mindset, but semantic searching opens the door to really find what you’re looking for quickly and efficiently without sacrificing quality.
Keep an eye out for our InnovationQ Tip post on using the semantic engine to find what you’re really looking for faster.