Monthly Archive: January 2019

New Version of Espacenet

A few weeks ago, the European Patent Office announced the beta release of the new version of its Espacenet patent search tool.  This post is a brief overview of Espacenet itself and of the beta release.


The beta release of Espacenet can be accessed here:

The current version can be accessed here:



Originally released to the public on 19th October 1998, Espacenet celebrated its twentieth birthday last year.  Following in the pioneering footsteps of the United States Patent and Trademark Office (USPTO), the Japan Patent Office (JPO) and IBM, Espacenet was and is the European initiative to bring the largest freely available collection of patent documents, and the ability to search them, as a public service to the global community via the internet.


Data Coverage

Espacenet has the broadest coverage of patent data of any of the freely available patent research tools.  It covers over 100 million published documents from over 100 countries.  Indexed and searchable data fields include title, abstract, description and claims; applicants and inventors; priority, application and publication dates and numbers; IPC and CPC classifications; as well as images, citation data and legal status information.

The new version will continue to sit atop this complete data set, although in the current beta release not all data may be available and this should be borne in mind when using it.


Assessment of Beta Espacenet

My immediate impressions are of a clean, modern interface that improves upon the previous (current) version of the tool.  Some parts are not fully intuitive, however with a little patience and experimentation the features and functionality quickly become clear.  Once this relatively easy learning curve is overcome, the tool is very user-friendly.


Features and Functionality

Having used Espacenet extensively for many years, I find it a powerful and comprehensive patent research product aimed and suited for general patent searching by the public at large.  Exploring the beta version, I’m pleased to see that none of the features or functionality have been lost.  There are, in fact, some welcome improvements.  Also, there are a few areas which might benefit from re-evaluation.



There are various ways of searching databases including:

  1. Simple individual keyword searches (form-based)
  2. Complex, multi-field searches (form-based)
  3. Advanced command-line searches

All three approaches are available in the current version of Espacenet.  Whilst they are also available in Beta Espacenet, only simple keyword searches and advanced command-line searches are presently accessible directly from the primary launch page of the product.  One has to first conduct one of these search types and view the results before being able to access the multi-field search form (labelled “Advanced search”).

Having noted this however, once the Advanced Search option becomes available it is clearly significantly improved upon that available in the current Espacenet.  All of the search fields previously available are present, with some valuable additions such as further text fields and text field combinations.  A very welcome development is the ability to create (via the form) complex queries using not only Boolean operators, but proximity operators and nesting as well.  The display for these complex queries is also a very well-designed graphical display showing clearly how the query terms relate and enabling easy editing.

Source: European Patent Office. “Espacenet Beta”. Retrieved January, 2019, from:

My only request for further improvement would be to have this advanced search interface available directly from the home page.  In the current beta version, accessing this interface is a two-click process for professionals used to working with this type of interface, whereas those using either simple keyword searching or command-line searching can start working immediately from the home page.


Results List

The results list is a clean easy-to-read layout, with various display options, including an image-only display, which can be very useful when searching in certain technical fields.

One negative point is the abstracts (when selected to be shown) are always truncated – one has to view the entire record to read the entire abstract text.  I would recommend an option to expand the abstracts to full in the results list if desired, otherwise remove the abstracts all together.

Records can be checked to select only those of interest, and the list conveniently reduced to display that subset.  One downside is that the interface is currently lacking a select/deselect all function.

There are download, print and “add to My patents” options, but these are hidden behind a vertical ellipsis and so not immediately obvious.

One great feature, is that the results list now displays to an unlimited length, rather than the first 500 records as in the current Espacenet.  It also displays an accurate hit count.  This is very welcome for any level of user as there is limited use in a patent research product if one cannot examine all the results found.



A new feature, and again a welcome one, is the introduction of Filters.  Not only do these enable the user to drill down in their results sets, but they also provide first-level analytics.  This introduces users to the power of patent analytics and enables the identification of other significant search terms, such as key assignees, inventors and patent classifications with which to develop searches.

The filtering function also benefits from the ability to either “apply” or “exclude” the filter selection to the results set.  Although multiple filters can be applied, it appears these can only be applied one at a time, resulting in a multistep process for some activities.  A breadcrumb trail of filter options is also presented enabling the user to undo any filtering option easily, but again only one filter at a time.

Source: European Patent Office. “Espacenet Beta”. Retrieved January, 2019, from:


Document Display

On the desktop version of the application, the results list, filtering options and individual document display are shown in a clean 3-column layout.  By hiding the filter display results in a one third results list and two thirds document display which is very useful.

The layout of the document data is very clean with all the previous fields available:

  • Bibliographic data
  • Description
  • Claims
  • Drawings
  • Original document
  • Citations
  • Legal status
  • Patent family

Of note are the additions of a claims tree (in the claims display) and a clear link to the Common Citation Document of the IP5 (in the citations display).  Both features add a powerful extra level of functionality for understanding an IP right.

The claims tree enables a quick, interactive interface for understanding and navigating the claims structure of a patent to identify the independent and dependent claims.

The Common Citation Document is a real-time application from the IP5 (the patent offices of Europe, Japan, South Korea, China and the USA) enabling users to understand global interest in a patent and its technology through citation activity at these five offices.

Finally, the European Patent Office’s Patent Translate function (a collaboration between the EPO and Google) remains available.  It allows text translation between any of the 28 official languages of the EPO’s 38 member-states, plus Chinese, Japanese, Korean and Russian.

Source: European Patent Office. “Espacenet Beta”. Retrieved January, 2019, from:


Session and Document Management

The My Espacenet feature provides a session log (My queries) and single “folder” (My patents) for managing patents of interest.

My queries is a clean overview of the searches run, with the ability to rerun or delete individual queries.  It does however lack a multi-action function, meaning queries have to be deleted individually.

My patents has the same functionality as the results list, with the same limitations (no select all).  Also, there appears to be no way of clearing the list and emptying the folder for a new project.


Cross-Device Responsiveness

The beta version is responsive across different device types.  Having tested it on a smartphone it maintains a positive user experience and retains the necessary functionality.


Overall Assessment

This is a beta version and the EPO acknowledges that the functionality, data handling or coverage are not yet at production level.  Nonetheless I found this a very positive and welcome development of a powerful and useful tool.

The current power of Espacenet is retained and developed to a new level with a modern, clean interface.  I would expect the beta version (when released to production) to do the job very well for basic (and more advanced) patent searching.  It is a great first step to accessing the wealth of patent information freely available online for initial patent searches and as a way to support your business and ideas strategy, growth and development.  For any business, organisation or individual, whether familiar with Espacenet or not, I would recommend exploring the tool and providing feedback with your own views and requirements to the EPO (there is a feedback link associated with the beta version).

Alistair Curson



European Patent Office. (2017, April 26). “Espacenet: free access to over 100 million patent documents”. Retrieved January 14, 2019, from:

European Patent Office. (2017, April 26). “Full-text (Databases and search)”. Retrieved January 14, 2019, from:

European Patent Office. (2017, April 26). “The worldwide patent database”. Retrieved January 14, 2019, from:

European Patent Office. (2018, July 19). “Patent Translate”. Retrieved January 15, 2019, from:

European Patent Office. (2018, November). “Welcome to Espacenet: free access to over 100 million patent documents”. Retrieved January 14, 2019, from:

European Patent Office. (2018, December 20). “Data”. Retrieved January 14, 2019, from:

Five IP Offices. “Common Citation Document (CCD).”. Retrieved January 15, 2019, from:

Patent Information News. (2018, September). “20 years ago: 1998: the launch of Espacenet”. Retrieved January 14, 2019, from:$File/patent_information_news_0318_en.pdf

Patent Information News. (2018, September). “Espacenet: the product that revolutionised access to patent information”. Retrieved January 14, 2019, from:$File/patent_information_news_0318_en.pdf

Artificial Intelligence and Innovation

Artificial intelligence (AI) is a rapidly developing field and technology space that has the potential to affect and disrupt innovation and business.  It even has the potential to generate new innovations itself.  In this post, I want to explore the capacity of AI to create innovations and how the current intellectual property landscape might view such creations.

What is Intellectual Property?

Intellectual property (IP) refers to the output of creative effort, i.e. “intellect”.  It is also, in law, a category of “property”, that can be owned, bought, sold, licensed and mortgaged.  IP is property that legally protects, particularly in the field of commerce, the creations of intellect as well as commercial reputation and goodwill.

Creators are able to earn recognition and / or financial benefit from what they invent or originate.  The law seeks to maintain a balance between the interests of those creators and the wider public interest.  The goal is to support an environment where innovation and creativity are encouraged and can thrive.

IP is often referred to in the context of human intellect or creations of the human mind.  However, as AI advances, at what point do we consider the creative effort of this type of intelligence and how does it fit into the current IP framework?

What is Artificial Intelligence?

Artificial intelligence (AI) describes intelligence within machines.  It is based on the premise that human intelligence can be defined in such a way that it can be mimicked by a machine.  Such machines are programmed to “think” like a human; to process data, learn, rationalise and solve problems; and to respond in ways that a human might.

Various levels of AI are often considered.  At one end of the spectrum are the purely reactive systems which respond solely to current experiences or input, and don’t capture or store memories or experiences for future reference.  At the other end are the self-aware systems, which perceive their own internal states, and can make inferences and formulate ideas.  In between are a range of systems with greater or lesser abilities to store and process information, make decisions, perform actions, and interact with people.

AI works by taking and combining vast amounts of data, iteratively processing the information at high speed, and progressively learning automatically from patterns and features it finds within the data.  It automates the learning process to manage quantities of data, as well as levels of analysis and accuracy, that a single human would find challenging, enabling it to get more out of the data.  However, it still requires a human to set up the system and ask the right question.

Applications of AI

As AI technology continues to grow and develop, it is finding applications in a wide range of everyday activities.  For example, in the healthcare space AI is being applied in the field of personalised medicine, as well as providing interactive personal healthcare assistants.  In retail its applications range from personalised shopping to stock management and designing the layout of retail space.  AI is used to optimise factory operations in manufacturing, and in sport it has a place in analysing images to help coaches improve the game played by their teams.

It’s important to remember that AI is not intended to replace people.  Rather its place is to augment our abilities and make us better, and we should work with it with that goal in mind.  AI adds a new layer of intelligence to existing systems.  For example, AI can improve upon existing analytical technologies, as well as bring analytics to industries and fields where such capability doesn’t exist or is underutilised.  It could improve our human abilities (e.g. vision, understanding and memory), and make us better at what we do.  Socially, it can break down barriers, such as economic and language barriers, leading to improved exchange and growth.

Potential for AI to Create IP

What is the potential for artificial intelligence to create intellectual property?  I want to consider two types of IP rights – patents and copyright.

Patents protect inventions – products and processes that are novel, inventive and capable of industrial application.  Could AI create an invention that satisfies these qualifications, and thus itself be considered an inventor?

Copyright protects works – an original literary, dramatic, musical or artistic piece; sound recording, film or broadcast; or typographical arrangement.  Could AI originate a work (fixed in a tangible medium) that qualifies as one of these subject matters, and thus itself be considered an author?

AI as the Inventor of a Patentable Invention

Under the New Zealand Patents Act 2013, an invention is defined as novel if it does not form part of the prior art base.  I discuss above how some levels of AI are defined by their ability to solve problems.  In this context, if an artificial intelligence creates something new, that qualifies as patentable subject matter, and that cannot be found in the prior art base (i.e. is not in the public domain), then the condition for novelty would likely be satisfied.

Similarly, for usefulness (“a specific, credible, and substantial utility” – New Zealand Patents Act 2013) or industrial applicability (“can be made or used in any kind of industry, including agriculture” – UK Patents Act 1977), it is likely that an invention created by AI could meet this criterium also.

Inventive step, on the other hand, is potentially a more subjective determination, despite established legal precedent as to its interpretation.  Under the New Zealand Patents Act 2013, an invention “involves an inventive step if it is not obvious to a person skilled in the art, having regard to any matter which forms part of the prior art base”.  This would likely be tested by evaluation of persons skilled in the art.  Assuming no bias was introduced by the witnesses knowing the inventor was an artificial intelligence, this test too could potentially be passed.

By these arguments, it is conceivable that an artificial intelligence could create an invention that met the criteria for patentability.  However, would such an AI be accepted as and consider an inventor?

Under the New Zealand Patents Act 2013, “a patent may only be granted to a person” who meets certain criteria with respect to the inventor of an invention.  Depending upon the definition of a person under New Zealand law, AI may not qualify as an inventor here.

The UK Patents Act 1977 is a little more interesting.  It states that “a patent for an invention may be granted primarily to the inventor” and defines an inventor as “the actual deviser of the invention”.  There may be scope for this to interpreted as the inventor not necessarily being a person.

These two examples highlight the jurisdictional variation that one is likely to encounter when exploring this question.  However, it does suggest that in certain jurisdictions AI may qualify as an inventor, even if clarification as to the interpretation of the relevant legislation might need to be sought.  The question of subsequent ownership of AI-inventions is one I’ll leave for a separate blog post.

AI as the Author of a Work in Which Copyright Subsists

Under the New Zealand Copyright Act 1994, a clear description of the works in which copyright would subsist are defined, as is the requirement for the works to be recorded.  Further, they should not be a copy of another work.  Taking these requirements alone, should an artificial intelligence independently (without copying) create and record one of the defined works, then there is an argument for it being the originator of the work, but would it be considered the author?

The New Zealand Copyright Act 1994 states “the author of a work is the person who creates it”.  Similarly, the UK Copyright, Designs and Patents Act 1988 sees the author as the “person” who creates the work.  Further, it defines the author of computer-generated literary, dramatic, musical or artistic works as “the person by whom the arrangements necessary for the creation of the work are undertaken”.  Examination of these two jurisdictions would indicate that AI is unlikely to be accepted as the author of a copyright-protected work.

It has been argued however (in relation to the UK Copyright, Designs and Patents Act 1988) that when this legislation was drafted there was a direct relationship or link between the input to a computer programme and the consequent output.  One of the functions of AI (as discussed above) is to analyse larger and sometimes unexpected sources of data than a human could thereby generating more complex and otherwise unpredictable outputs.  This opens the door to discussion as to the true nature of creativity.

Existing case law in the US and Germany for example also establishes precedent for copyright to subsist only in works created by humans.  However, as the complexity of AI and its abilities develop, it is not inconceivable for future precedent where AI becomes accepted as the originator of copyright-protected works.


The current definitions of AI are many and complex, and these may well develop and expand in the future.  There appears already to be scope for certain IP rights to be legitimately recognised as created by AI.  In the future, our legislative framework may move towards recognising the creative abilities of AI even further.

I haven’t touched upon ownership of the relevant IP rights here, but I do think this needs to be part of the overall discussion, in particular within the context of the motivations behind IP rights and their protection – the incentive of rewarding creators for the fruits of their creations, protecting investment and thus incentivising innovation.

Fitting this into the framework of IP legislation is likely to prove challenging.  Beyond who or what created the IP, the ownership and enforcement of the rights is a key part of the equation.  With AI-created innovations will those incentives remain and if so, in what form?  Maybe society will need a new form of IP right or rights to recognise and manage those innovations created by AI.  I do think that AI will continue to develop and become more and more integrated into society and our lives.  It will inevitably become a source of creativity and the progression of intellectual effort and this will need to be managed.

Alistair Curson



Bainbridge D. (2012). Intellectual Property, 9th edn, Pearson Education Limited, Harlow, England.

Bakry MF, He Z. (2015, March 5). “Autonomous Creation – Creation by Robots: Who Owns the IP Rights?”. Retrieved November 29, 2018, from:

Chisling A. “Enter the Dawn of Artificial Intelligence and Intellectual Property”. Retrieved November 29, 2018, from:

Desai A. (2017, October 7). “What are the Types of Artificial Intelligence”. Retrieved November 28, 2018, from:

Investopedia. “Artificial Intelligence – AI”. Retrieved November 28, 2018, from:

Macaulay T. (2018, January 26). “Legal Issues Around IP For AI: Who Owns the Copyright on Content Created by Machines”. Retrieved November 29, 2018, from:

Medeiros M, Sanft J. (2018, January). “Artificial Intelligence and Intellectual Property Considerations”. Retrieved November 29, 2018, from:

Menon N. “Who Owns the Content Created by an AI? Patent and Copyright Privileges”. Retrieved November 29, 2018, from:

New Zealand Copyright Act 1994, s5. Retrieved January 4, 2019, from:

New Zealand Copyright Act 1994, s14. Retrieved January 4, 2019, from:

New Zealand Copyright Act 1994, s15. Retrieved January 4, 2019, from:

New Zealand Patents Act 2013, s6. Retrieved December 6, 2018, from:

New Zealand Patents Act 2013, s7. Retrieved December 6, 2018, from:

New Zealand Patents Act 2013, s10. Retrieved December 6, 2018, from:

New Zealand Patents Act 2013, s22. Retrieved December 6, 2018, from:

Penner MD. Vanderveken M. (2018, June 27). “Could the Creations of Artificial Intelligence Be Entitled to Intellectual Property Protection? Part 1: Could Artificial Intelligence be Considered an Inventor?”. Retrieved November 29, 2018, from:

Penner MD. Vanderveken M. (2018, July 10). “Could the Creations of Artificial Intelligence Be Entitled to Intellectual Property Protection? Part 2: Can Artificial Intelligence be an Author According to Copyright?”. Retrieved November 29, 2018, from:

SAS. “Artificial Intelligence: What It Is and Why It Matters”. Retrieved November 28, 2018, from:

United Kingdom Copyright, Designs and Patents Act 1988, s9. Retrieved January 4, 2019, from:

United Kingdom Patents Act 1977, s4. Retrieved December 6, 2018, from:

United Kingdom Patents Act 1977, s7. Retrieved December 6, 2018, from:

Wikipedia. (2018, November 23). “Intellectual Property”. Retrieved November 28, 2018, from:

World Intellectual Property Organisation. “What is Intellectual Property?”. Retrieved November 28, 2018, from: