Problem with PDF Documents Filed with IP Australia
08/12/17
Greg Gurr, Convenor, IPTA IP Australia Business Relations Committee
Problem with PDF Documents Filed with IP Australia
IPTA has identified a problem with PDF documents, including patent specifications, filed with IP Australia as a result of IP Australia’s processing of incoming PDF documents. The processing by IP Australia has resulted in corruption of some documents, with the stripping out of various special “Unicode” characters from documents.
IPTA has been liaising with IP Australia for some time, seeking to gauge the extent of the problem and prompting IP Australia to take appropriate corrective action, and now wants to update Members on this situation.
Details of what IPTA currently understands of this problem and corrective measures taken (or being taken) by IP Australia are set out below.
Background
All PDF documents filed with IP Australia via e-Services and B2B are processed with Ghostscript document conversion software to convert the PDF documents into a standard format for loading into IP Australia’s systems. All text and symbols used in electronic writing systems are identifiable by a computing industry standard encoded system known as “Unicode”. The PDF conversion process used by IP Australia relies on the Ghostscipt software identifying all Unicode characters in any document being converted.
Identification of the problem
IPTA became aware earlier this year of instances where documents on file at IP Australia did not match the documents actually filed via e-Services or B2B, with various non-text symbols being omitted from the document in IP Australia’s electronic record.
By way of example (and emphasising that this issue is not confined to chemical subject matter), the arrow in the following equation was stripped out of a patent specification filed using eServices.
After investigation by IP Australia, it was determined that there have been failures with the Ghostscript document conversion process whereby various Unicode characters in PDF documents have not been identified. When a character cannot be recognised, it seems that the character is simply stripped out and thus omitted in the converted form of the document that is loaded into IP Australia’s systems. The version of Ghostscript that IP Australia had been using was not recognising some of the more unusual characters, particularly various non-text symbols. IP Australia has unfortunately not been able to identify the specific characters affected, nor have they been able to obtain any detail from the authors of Ghostscript.
IP Australia believes the issue could affect all documents filed since the commissioning of e-Services and B2B in 2011 up until April 2017.
Solution
After IPTA alerted IP Australia to the problem early this year, IP Australia updated the Ghostscript conversion software in April 2017. IPTA has been assured that this software update has resolved the majority of circumstances where conversion errors were occurring. IP Australia has, however, advised that there are still some bugs that they are still investigating. IP Australia is also internally reviewing their reasoning for converting all PDF documents and investigating the possibility of removing or modifying the process to skip some of the document conversion steps.
Steps being taken to identify corrupted documents
For quite some time IPTA has been pressing IP Australia to identify all documents that might have been corrupted, highlighting the potential impact and urging the allocation of resources to fully resolve the problem. However at this stage, it does not seem that there are any realistic solutions to identify all documents that may be affected.
IP Australia has, however, put in place examination processes whereby Examiners are required to look out for documents that appear to have special characters missing. When an issue is identified, the Examiners can access the originally filed documents (i.e. those submitted via e-Services or B2B) and replace the corrupted documents with the originally filed documents.
Unfortunately it is also difficult for Members to identify corrupted documents, unless they compare documents as filed from their own systems against documents available in the case’s e-Dossier on AusPat.
Members are, however, encouraged to be aware of this problem, particularly when prosecuting applications with formulae or other special characters, and to compare their own filed documents against those in the application’s e-Dossier on AusPat.
If any Members do come across documents that are corrupted, please alert IPTA so that we can monitor the extent of the problem and continue working with IP Australia to better characterise the problem.
Greg Gurr
Convenor – IP Australia Business Relations Committee
December 2017
