As long-time readers will be aware, it was almost exactly six years ago that I announced that I was leaving my job as a patent attorney with Watermark (which has since disappeared after being ‘integrated’ into Griffith Hack) to pursue other, vaguely articulated, interests. Since then I have operated as an independent IP consultant. I have undertaken some ‘traditional’ patent attorney tasks, including patent drafting and prosecution, continuing to work with some of my colleagues and clients from the Watermark days, as well as a few new ones along the way. But I have also had the opportunity to engage in a range of projects that simply would not have been open to me within an attorney firm environment, including working with IP Australia on the IP Government Open Data (IPGOD) releases in 2018 and 2019, assisting a university research group with patent mapping and developing/documenting an IP strategy for a major grant application, and consulting to investment analysts on the Australian market for IP services.
Some of the most interesting work I have been involved in over the past few years has come about as a result of the software tools that I have developed for processing and analysing Australian (and New Zealand) patent data, and the many articles I have published here making use of those tools. I have also dabbled in machine learning (ML) and natural language processing (NLP) technology, using techniques from these fields for ‘fuzzy’ matching of entity names in patent data and for distinguishing between individual and corporate names in unstructured data (both of which were employed in my work on IPGOD, and are in use every day in my automated patent data updates), as well as in more speculative applications such as classifying provisional applications into technology fields based only on published titles.
A couple of years ago I started thinking about some important, but difficult, problems in patent law and policy that might be addressed using some of these kinds of technology. For example, what if we want to measure the effect of different patent laws and examination processes in different jurisdictions? The Australian government’s Productivity Commission thought this was a relevant consideration in its 2016 report on the nation’s intellectual property arrangements, recommending (among other things) that the legal test for inventive step should be modified (yet again – after it had only been adjusted in 2013) to bring Australia more closely into line with Europe. IP Australia’s initial efforts to implement this recommendation have been abortive, and with the recent change in government seem likely to fall by the wayside altogether.
