gremlin / iStockphoto.com
Octimine’s natural language processing tools have the power to transform clunky patent database queries into quickfire searches, according to Steven Shape of Dennemeyer & Associates.
Patent databases worldwide are multiplying. In 2018, 3.3 million patent applications were filed at patent offices worldwide, a 5.2% increase compared to the previous year, according to the World Intellectual Property Organization (WIPO).
This surge is mostly due to China’s arrival as a significant source of patenting activity. In 2018, Chinese patent applications comprised nearly half of all patent application filings that year.
With about 120 million patent documents sitting in databases across the world, many IP professionals face considerable challenges traversing the broad landscape of existing patent literature when trying to conduct a prior art search.
The high level of technical detail in each patent filing makes it difficult to find the one or two essential pieces of prior art that might exist among the thousands of patent filings in a particular technical sector.
Patent searches are made more difficult by the different classification systems in use across the world. A keyword search may not be as complete if the prior art you seek exists in a separate patent classification from the one your search is targeting.
How Octimine makes prior art searches more efficient
Although many patent database searching tools exist on the internet, we feel that few improved search tools match Dennemeyer Octimine’s sophisticated search technology. With artificial intelligence (AI) and machine learning features built into its search engine, this software solution enables semantic patent searches based on natural language text inputs.
Octimine, acquired by Dennemeyer in 2018, provides an easy-to-use interface so the user does not have to remember search syntax or use proper Boolean operators (‘and’, ‘not’, and ‘or’) to get results. Octimine drastically reduces the amount of time needed to perform a prior art search, and its intuitive design also opens up potential new uses of patent searches to promote innovation and creativity.
Octimine’s co-founder Matthias Pötzl first encountered ineffective patent database searching while he was a student carrying out work for Siemens. Despite using the patent search tools made available by the US Patent and Trademark Office (USPTO), Pötzl failed to find the results he was seeking.
While pursuing his PhD, Pötzl met Octimine’s other co-founder Michael Natterer, who was focusing his own PhD research on patent data analysis. The algorithms developed by the pair have been tested with large datasets enabling the engine to build a kind of “fingerprint” from natural language text inputs.
Those “fingerprints” are compared with existing ones that have already been calculated for patent database documents.
“We have trained the machine on a certain pattern, and it tries to find this multidimensional pattern in a big database to retrieve similar documents,” Pötzl says. “It can replicate patent examiners’ work in finding relevant prior art references.”
AI is necessary to process natural language
Many of the legacy patent search tools available are built on database platforms using computer codes that are able to process natural language. The human brain is wired to notice clues communicated at a high level of abstraction eg, understanding that a sarcastic comment suggests a viewpoint that undercuts the surface level meaning of a speaker’s words.
In contrast, many common programming languages do not have the capability for abstract semantic analysis. They derive meaning from search inputs with rigid low-level processing rules. In the context of patent databases, search tools based on these simpler programming languages often require users to master confusing syntax operators to filter search results.
At best, such filters refine a user’s search results only coarsely, forcing users to manually review those results and develop new iterations of a search to uncover better results.
The new generation of search engines, of which Octimine is an example, incorporate various forms of machine learning into their software to retrieve search results tailored to work well with natural language inputs.
Anyone who has used Google to directly canvass the internet with a question, rather than search terms carefully plotted with Boolean operators, has experienced more targeted results courtesy of machine-learning algorithms designed to process natural language.
The AI and machine-learning algorithms developed for Octimine give users the ability to find relevant search results by copying and pasting text from technical articles, Wikipedia or even product descriptions on shopping sites.
While increasing the ease of conducting a patent search using natural language inputs is an obvious benefit of the AI technology incorporated into Octimine, it is hard to exaggerate the impact of AI on getting more useful search results. Octimine returns highly relevant results and extremely useful visualisations, allowing for more in-depth analysis of patent documents that are similar to the search input.
Along with the traditional search filters typical of patent databases, Octimine’s visualisations aid users in gaining a greater understanding of a technology’s life cycle, economic impact and legal risk for particular patents.
The software also provides citation and reference maps so the relationship between similar patent documents and patent families can be seen clearly.
A precious tool
When used together with traditional patent searches that use Boolean operators, Octimine can retrieve useful documents in a fraction of the time it takes to find those same results through multiple iterations of different search-term combinations.
Pötzl says: “IP professionals can do manual patent searches and then feed Octimine with the results to see if the machine returns new documents that they may have missed. It may be the case that one document sounds different, but it covers the same technology. Users can then adapt their keywords and improve the quality of their searches.”
Octimine subscribers can complement the use of the AI software with traditional Boolean-based searches eg, the Swiss Patent Office has implemented Octimine as a standard tool, requiring its use for every Boolean search conducted through the office’s search tools.
Many companies choose to outsource IP searches due to their time-intensive nature. A typical prior art patent search could take an entire work day, and the expense of billable hours means these searches are often relegated to whoever is available.
Even law firms large enough to establish patent divisions may only have an employee or two available to work on prior art searches. For multinational enterprises with more extensive patent portfolios, patent searches can become very expensive and they may not pinpoint every vital document, leading to the unfortunate discovery of relevant prior art by an examiner during patent prosecution.
Octimine’s powerful algorithms are valuable beyond the confines of patent prosecution. Firms that want to challenge the validity of existing patents, especially those that might be asserted against them in litigation, can use Octimine to reduce the cost of engaging in such proceedings by quickly finding prior art relevant to obviousness or anticipation challenges.
Pötzl says: “In some cases, our customers were able to find the ‘silver bullet’ prior art reference capable of invalidating a patent in a very short time.”
While validity challenges are always a possible complication in major patent jurisdictions, including the US and Europe, Octimine can help save time, money and mental labour for technology companies.
In many nations, especially throughout Europe, industrial corporations often encourage their research and development (R&D) staff to review existing patent literature in a particular sector to stay on top of recent developments.
“For R&D professionals, it can be advantageous to look at what is going on with patenting at a competitor. They can use it as inspiration for new ideas on how to solve a problem,” Pötzl explains.
With Octimine, such inspiration can be found without spending most of the day on composing complex keyword searches to uncover every useful resource.
Even in areas of the world where legal regimes encourage R&D professionals to sandbox their knowledge of relevant patent literature to ward off allegations of wilful patent infringement, the knowledge economy can still benefit from Octimine’s patent search and analysis tools.
In countries such as the US, where many technology companies try to dissuade R&D professionals from reviewing granted patents and patent applications, engineers who are able to review patent literature become more aware of cutting-edge solutions.
This access to the existing documentation can promote ingenuity among engineers, who then find new solutions to technological problems without infringing on existing patent rights. This approach encourages scientific progress, a foundational justification for the US patent system’s existence.
Growing its global presence
Octimine is not the only company offering patent database search tools that use AI technologies. However, it is a cost-effective alternative to existing platforms that provide subscribers with highly developed patent search capabilities.
Other firms with similar software-as-a-service (SaaS) cloud platforms, devoted to the analysis of patent literature and entire corporate patent portfolios, have annual subscription rates costing tens of thousands of dollars. Furthermore, some of those expensive solutions have different uses that are best suited to promoting innovation life cycles, rather than performing thorough surveys of existing technical literature.
With Octimine, a single account to gain access to the AI patent search platform starts below $3,000. This allows companies to buy multiple seats and quickly gain a handle on the existing patent literature to positively influence their business and R&D activities.
No technology field remains stagnant for long, and Octimine’s development team continues to expand and improve upon the software. New features on Octimine include a slider mechanism, which allows users to emphasise components of textual inputs to retrieve search results more closely related to them.
Pötzl believes that future developments on the Octimine platform will support monitoring and alerting mechanisms that allow companies to stay up to date on patent literature in particular fields.
Octimine is a popular SaaS patent search solution in Europe, thanks in part to its origins at Germany’s Max Planck Institute, giving the platform an advantage in that market, but although Octimine can already claim several significant US companies and institutions as users, the patent search engine is still a novelty in the region.
Now working as a part of the Dennemeyer IP Group with a global corporate infrastructure in place, the Octimine brand is set to expand globally as a broader business community begins to recognise its benefits.
Steven Shape is a managing partner at Dennemeyer & Associates. He is a registered patent attorney and electrical engineer with extensive experience in all aspects of Federal Court litigation. He is an adviser to many Fortune 100 companies in the US and worldwide. He can be contacted at firstname.lastname@example.org.
Octimine, Dennemeyer, databases, machine, operators, Boolean, companies, technology, natural, platforms, software, patent