How to perform a validity search part 2

In part 1 we discussed that a validity search order specification should include as much information as possible, such as the reason for the search, which claims or features should be examined, what information you have regarding office actions or communications, etc. This provides an optimal starting point and focus is directed to what the client needs for their case and argumentation.

Now it is time to dig into the part of “how to perform an optimal search”. A validity search is a bit like an Escape room, where the information specialist “travels back in time”. The information specialist starts in Room 1, “the debriefing room”, to get a rundown of the preconditions and time frame for the search, and then continues to Room 2, “the patent search room”. The similarity to an Escape room is not only the time travel but also the creative process of finding trails that can lead to another trail, and then defining the next step in the process based on what trails are found. A search performed by a specialist is difficult to define, since it is not a process with specific steps, it is more of a creative process that can take different routes based on findings during a search. An information specialist is always trying to follow the trail of breadcrumbs, and when doing so, the search can take a different route.

Entering room 2, “The patent search room”, we enter a clean and well-arranged room, similar to the patent data, that is well structured and with several search possibilities. Here we take some time to consider what toolbox was available for the patent examiners and to identify any possible new tools that may be available to us, for example:

New patent classes can provide a more targeted search than the examiners’ searches, such as CPC/IPC/F-term classes that were not previously available, or CPC classes such as indexing codes that have been added to classes that have grown.

  • Similarity searches can combine text and metadata searches. Text searches are based on traditional Boolean searches with keyword combinations or algorithms and are usually named AI, Semantic etc., depending on the supplier. The metadata is normally a combination of classes and citation trees using weighting factors based on for example relevance of citations/citing, if classes are CPC or IPC, and classes that are marked as most relevant. Additional metadata that can be used is e.g. inventor names and applicant names.
  • Terminology: when large or abrupt technical changes arise, the terminology will not be settled for a few years. For the information specialist to find the “undefined terminology” at the time of filing an application, the search may also be done for later documents. Relevant documents found can be used to go step by step back in time, based on their cited documents, all the way to the time of filing, and then the terminology used at that time can be identified.
  • AI: Without trying to dive into what AI is, or is not, and how suppliers define their tools, it is important to remember that such tools were not available for the examiners. We can use them to search directly on the opposed patent with different inputs such as the whole patent, only claims, title and abstract, and so on. These approaches must be considered since the search results can be very different depending on the contents of the description, for example if the background technology forms a large part of the document, or in a divisional application where several different embodiments are described but only one is included in the claims.
  • Languages: improved machine translations increase the possibility to retrieve relevant documents that were not found by the examiners. Another approach is to search in original languages especially in case the investigated patent does not have any family members in countries where interesting corporations and research institutes reside. This provides a search in a language that has not been searched before by a patent office. The search in original languages can retrieve documents that were not translated correctly by machine translations which can be the case for documents that were only published in non-english languages.
  • Time: how much time does an examiner use for a search? Consider what could have been possible to search within that time frame and then broaden the search to include more classes and other technical fields. As some colleagues say, sometimes you must use brute force and just review all documents in some classes.

So, as well as approaching a validity search as an ordinary search, we also consider everything that may add something new compared to the examiners’ tools. The approach is selected based on experience and the search can follow different paths based on what is found along the way. To summarize, a search is a creative process, and we call it “The art of finding”.


Did you miss part 1 of „How to perform a validity search“? Click here.