Patent Invalidity Search: New Approaches to Prior Art Search
What is Patent Invalidity Search?
A patent invalidity search is a process to invalidate a patent. The patent invalidity search is performed to find the relevant published literature (including patents, technical papers, products, manuals, and even university theses), videos, and/or products available before the priority/filing date of a patent.
1. For new inventions, a prior art search helps you determine how likely is it that your invention (or portions thereof) will receive a patent.
2. For existing patents, a prior art search helps you challenge their validity (for example, through an Inter Partes Review at the Patent Trial and Appellate Board).
3. For patent licensing, a prior art search is performed to extract high-value patents in a patent portfolio and explore licensing opportunities. (Patent portfolio mining)
As per the statistics from 2015, almost 90% of petitions instituted at the PTAB result in at least one claim being invalidated – while 20% of the petitions result in a full invalidation of all asserted claims.
80% of PTAB petitions that are fully instituted result in successful patent invalidity – and the rate drops to almost 50% for partially instituted petitions. This statistic highlights an important point – a deep and diligent patent invalidity effort at the institution stage greatly increases the success rate for the petitioners. Even outside the realms of the PTAB, patent invalidity through §102 and §103 prior art issues remains the most popular and successful strategy for defense.
Finding and analyzing the right prior art at the early stage can be difficult, of course – though a number of consulting outfits are available as an inexpensive resource for lowering the costs of the search. Streamlined processes and economies of scale and geography generally allow such firms to reduce the cost of prior art searches by over 40-60% (compared to using law firm resources) – and bring down the total cost to $1000-$4000 range per patent. However, no matter the price, quality and thoroughness by such outsourcing vendors remains circumspect. Generally, most vendors employ one or more of the leading patent databases (such as Thomson Innovation and Innography) and a few non-patent literature databases such as ACM and Citeseer. These databases provide good coverage on potential sources of prior art – yet, gaps remain that limit their success rate, especially in complex technology areas such as pharmaceuticals, semiconductors and material sciences.
In a quest to provide maximum value to clients at minimal cost, most prior art search providers rely almost exclusively on a keyword-based approach, and do little else if they do not find good prior art from that approach. In this paper we explore several additional strategies that can help attorneys delve deeper to find the right prior art.
New Approaches to Find the Right Prior Art
1. Citation Traversal
Analyzing backward citations of the asserted patent is important – yet seldom useful for the purposes of patent invalidity. By their own virtue, the backward citations have typically already been considered by the inventors and the examiner during file action. However, backward citations as well as forward citations can reveal other prior art on a deeper second-level analysis:
Backward Citations of Backward Citations
Other Forward Citations of Backward Citations
Other Backward Citations of Forward Citations
Similarly, the search can be expanded further to a third level as well using the same approach – where keywords have failed to deliver the right prior art.
2. Inventor Traversal
Seemingly a logical step in the prior art search project, an inventor-based search is often overlooked by attorneys and engineers engaged on a prior art search. The inventors named on the asserted patent are more likely, even individually, than others to have disclosed prior art in the years preceding the asserted patents. Therefore, searchers should pay special attention to the corpus of literature attributable to each of the named inventors.
Further, inventor-based analysis should ideally be combined with the citation traversal – where other work attributable to the inventors named on backward citations and forward citations, as well as on second level citations enumerated above, should be considered in particular.
3. Patent Class Traversal
Most prior art searches begin with a keyword search – where the keywords are iteratively refined in order to arrive at a manageable set of 20-50 top results that the searchers then review manually in order to ascertain the true relevance to the asserted claims. To reach this manageable set, searchers would sometimes act conservatively and eliminate some of the synonyms or prerequisite keywords to filter out unlikely candidates. Unfortunately this conservative approach still results in a number of false positives in that final set of 20-50 results – requiring additional iterations (and effort).
Identifying key patent classes (preferably, IPC classifications due to their global applicability) and limiting patent search to those classes increases the odds of finding the best prior art within that final set. Identifying the key classes can however be tricky – requiring diligence than simply using the IPC classes listed on the asserted patent. The following approaches at least are recommended:
IPC classes of backward and forward citations
IPC classes of second-level citations
IPC classes of other patents by the same inventors
4. Japanese/Korean/Chinese Translations
While most patent databases including Thomson Innovation, Delphion and Innography include patents from Japan, Korea and China – they either contain only English-language abstracts or at their very best machine translated versions of the patents.
These jurisdictions are a treasure trove of prior art in the high-tech space. Japanese patents for example far exceed the total number of patents filed with the USPTO – with an overwhelming majority of them filed by residents. Yet, traditional approaches to prior art search usually deliver an underwhelming number of Japanese results – which means that traditional prior art search methods do not adequately search within these jurisdictions – presumably suffering from limitations inherent to machine translation.
If the particular technology area you are searching is known to have a lot of innovation from one of these countries – for example, consumer electronics and semiconductors – a native-language search is highly recommended. Machine translations are often adequate for a general understanding of the invention – but only a native speaker would be able to search patents with due consideration to dialects, local terms and conventions – as well as analyze intricacies within the results in deeper detail.
5. Historical Search
Technology moves at an alarmingly fast rate today – and innovative companies, especially small and medium enterprises, go out of business or get assimilated into larger entities all the time. When that happens, it is not uncommon that public information, marketing material and product documentation is taken offline. Unless their technology was widely popular or were widely patented, documentation on it can vanish from the public eye entirely.
Fortunately however, Internet has a time machine –Archive.org. Archive.org periodically crawls the Internet and captures snapshots of websites – such that even if a website is subsequently taken down, a user can later view the website content as it existed at the time of crawling. For most website, Archive.org stores dozens of dated copies.
Archive.org is however, not a search engine – and does not allow users to search for website content a la Google. However, an expert with some experience in the related technology from the time of the asserted patent should be able to guide the searchers to look for specific companies that may have marketed prior art products at the time. Archive.org then serves as a useful resource for exploring documentation on those discontinued prior art products.
Prior art search is often an open ended endeavor with an unpredictable success rate. It can take anywhere from 1 hour to 100 hours to find adequate references – but using creative approaches including the ones listed herein, beyond keyword-based searches, can help increase the success rate notably.