IPR&D Information Adventurers Blog

Patent Search Strategy: Creating the Perfect Search

Before we consider what constitutes a good search, we should ask if there is such a thing as “the perfect search”. Could one generate a novelty or validity search that finds absolutely all the relevant documents?

In the normal time allotted to search creation, it is probably not possible: there are too many potential documents and too many different search strategies to test. It’s highly unlikely that under normal circumstances, one could find all relevant documents.

That said, I have once seen a perfect search executed, although it certainly wasn’t under normal circumstances! However, the story should be told, because there’s an important lesson related to good IP and R&D searches. I certainly learned something that we apply at Evalueserve for our own processes.

It is a reality in our still fragmented industry that any head of IP at any large corporate entity receives many unsolicited calls every week from patent search providers looking for commissions.

One head of IP decided to put this reality to work for them. They sent the same validity search request to around 40 reputable patent search providers that had pitched their services to the company. The request in each case was that the search provider should do the search as a free pilot.

In case you’re not familiar with the search type, a validity search has the objective of identifying any prior art that might invalidate a patent. At this point in the prosecution process, patent offices have already performed their searches and found a good selection of obvious documents. Therefore, a validity search must dig deeper than the type of novelty search that is normally done.

The 40 search providers clearly put their smartest people on the job, investing considerable time and resources, because the consolidated output, which I had the opportunity to see, was really amazing.

Most of the search providers had found some interesting new documents. What was particularly interesting is that many of them had identified one or two valuable results that were unique to their output. Thus, when consolidated, the results were essentially the perfect search, giving all the required information with only a de-duplication exercise needed to clean up the answer set.

I said this ‘perfect’ search is not repeatable. You’ll agree I’m sure that the ethics of the “commission” were questionable. What’s more, the effort is clearly not feasible in any company’s daily routine. But, as I mentioned, we can still learn two important from this.

First, if you give two search professionals freedom to come up with their own strategies, when you put the output together, you may be able to use the information from both strategies and the hits retrieved to further improve the combined output. This is a principle that we have implemented at Evalueserve: we are one of the world’s largest providers of IP and R&D services, with a pool of several hundred research and analytics specialists to draw from. When creating a search strategy, we draw on expertise from multiple professionals, not restricting ourselves to a single point of view.

Second, since the number of documents and the number of possible strategies is too large for most standard searches to produce a 100% retrieval, we should take this into account in the creation of a good search. A good search should always aim for as close to 100% retrieval as possible. How can we achieve this? And how can we determine or measure the quality of search. We will examine this more closely in the coming posts.

Have you ever seen a search that you considered perfect? Have you ever had the opportunity to see the combined output from multiple search strategies? Are there any other learnings we can gain from such combined search strategies? Let us know in the comments.

Speak your Mind

Your email address will not be published. Required fields are marked *

  1. Honey Garg says:

    I feel, the strategy from multiple or 2 searcher gives confidence to say “Search is completed”.
    But I have seen scenario’s when ever i was in loop with inventor, something new has come us giving me the feel of completeness.
    Definitely 2 or 3 minds come up with more permutation & combination.

    1. Evalueserve says:

      Thanks for the feedback and the confirmation that parallel searching or an individual choosing a methodology that uses ‘different‘ retrieval strategies will improve quality of the search. Very interesting feedback on the interaction with the inventor – we will discuss this critical role of ‘communication’ between search specialist and customer (inventor, patent counsel, etc.) in one of our upcoming blogs.

  2. Sumeet says:

    Its thunderous to know that someone is looking for definition of perfect search. Well, it is hard to pull up a strategy that can produce or deliver 100 percent accurate result. I saw an innovative approach where multiple strategies were made for single project. In fact I implemented on my own. If you think from inventor pint of view, it would certainly help. For better result, first go through approximate last five years of patents categorizing using CPC, UPC with some keywords. Moreover, if competitor of inventor is analyzed at first, we can go their patent portfolio. It would lessen our effort and hit the bulls eye.

    1. Evalueserve says:

      Thanks a lot, Sumeet. We look forward to continuing the exploration of search quality in upcoming posts. You are hitting the nail on the head – in daily routine situations we don’t have the luxury of unlimited resources, so we need to cast the net where the fish are and design initial search strings in areas where there is higher likelihood of success. For example, this includes starting with known assignees/inventors or limiting with smart keywords – we will discuss in upcoming posts not only how to clearly define the ‘bulls eye’ but also how this can be done in a systematic and transparent way so that the customer (inventor, patent counsel, etc.) understands how the search was done and that we have done the search in the most efficient way.

  3. Vipul Mendiratta says:

    While we are trying to define a perfect search, one must also consider the question of ‘purpose’ behind each search. From a consumer perspective, a search is perfect if the search uncovered a 102 reference for a Patentability and 100% retrieval may not be required. And same reference may not be a good from a validity standpoint. Here, the search strategy cannot be questioned because a 102 art was identified, but from a validity stand point may be all variation of embodiment were not identified and 100% retrieval is important. Hence, quality of a search may not be evaluated alone on the retrieval of a matching set, but the purpose is important too. Look forward to more updates in Part 2.

    1. Evalueserve says:

      Thank you very much for the feedback – you are absolutely right! The ‘purpose’ or ‘use case’ is key and will define what kind of quality is required for the specific patent search or patent landscape. We will consider this further in parts 3 and 4, where we not only introduce the concept of use case but also present a formula for the ‘search quality index’ that is applicable for all different ‘use cases’ in IP or R&D. Other factors, such as type of end user and resources available, will also play a role.

      You are further correct – 100% recall is not required for the case you describe – in that case, the job is done! However, there are other cases (e.g. Freedom to Operate Search) where the situation is different.