Data visualisation

Client has a workflow system that dumps 70,000 records about the status of their 10-step production process in a CVS file regularly. They wanted to convert all these data records into 16 pie charts, one for each division of their main customer, so that the account managers can keep track and adjust their planning. Their request was to create VBA code for Excel to automatically recreate the pie charts twice a day.

Problems: coding VBA for Excel is cumbersome, execution is very slow, presenting 16 pie charts to choose from might cause confusion and the customer's IT department would not allow Excel macros in the first place due to security concerns.

Solution: using an open source Javascript data visualisation library, we created a single pie chart with easy-to-use controls to select the division and products to be included. No transfer of Excel files to customers and instant updates from the production tracking system meant immediate and unproblematic insight for the clients. Coding time was less than 10% of what the VBA solution would have taken.

Post-translation DTP automation

Client is handling DTP for the Official Journal of the European Union, with volumes ranging from 1 to several 100 pages, published simultaneously in the 24 languages of the EU. Requirements include consistent styling across all language - making sure the same content shows on the same page in every language.

Problem: injecting page breaks to align content across all 24 languages is a tedious and labour-intensive job. Moreover, the employees who were doing this job usually did not understand the target language so they needed to count paragraphs and table rows to make sure the page breaks were in the right locations. Changes to styling in the model language could cause all of the tedious work to be repeated from scratch.

Solution: using a simplified XML schema specifically for styling, with a scripted user interface in Adobe FrameMaker allowed handling of style changes without the need to redo any pagination in any of the languages. A pagination script took over the work of aligning all paragraphs and table rows after all styling was done. A large part of styling was already handled in the original conversion to the specific XML. The process saved enormous amounts of time and allowed the customer to accept jobs with an 8-hour turnaround time.

Conversion to XML

Since the start of this century, practically every computer document is stored in one of the many XML standards. Older content, mostly written in MS Word, is known as unstructured content and requires conversion. Depending on the state of those older documents, it might be more cost-effective to rewrite everything in XML from scratch.

Problem: the customer had 100,000 documents, each covering a single page. About 60% was in MS Word and the remaining 40% in AutoCad (because of graphics on that page). All of this unstructured content needed to be converted to XML. Manually recreating the content was not a viable option.

Solution: all Word documents were read into Adobe FrameMaker and converted into XML using an author-guided self-learning script. All AutoCad files were read into Adobe Illustrator, with automated scripts extracting text content to XML files and isolating images in SVG files. Once all single pages were processed, an XSLT process glued everything together into a small number of XML files.

Making XML data readable

Client calibrates technical equipment using proprietary XML to store and export data. Conversion of the complex XML into organised tables was done, with the remaining parts of the test results documents written and maintained in DITA.

Problem: specifications for the proprietary XML changed, causing issues with the existing conversion to DITA. Also, the process to handle the generated DITA files and edit the remaining files was cumbersome and caused a bottleneck as only a few employees had access to and knowledge of the DITA publishing pipeline.

Solution: conversion of the proprietary XML into a set of tables was corrected for the XML changes and enhanced with introductory paragraphs plus layout to match the required calibration report layout and standard content. CSS for Print was used to make the resulting online pages printable with headers and page numbering. The entire DITA publishing process was skipped, allowing every employee to create their own calibration reports.

Troubleshooting

Troubleshooting topics in technical documents are usually pretty useless as they just mention the obvious. For large industrial installations this often means flying in an expert to assess the situation and figure out what the true cause of the problem is.

Problem: troubleshooting tables (showing symptoms, possible causes and remedies) do not solve complex problems which require a step-by-step ruling out of possible causes to finally reach a solution that works. Such expert knowledge is some kind of implicit flowchart, existing only in the mind of the expert engineer. Sending these experts to customers all over the planet is expensive and makes them unavailable for essential development work at home.

Solution: using BPMN with an open-source editor, flowcharts were created together with an SVG visualisation. The BPMN is converted to an interactive step-by-step procedure in HTML, along with the SVG showing the current step. Once the procedure is tested and found to be complete, the interactive procedure - without the SVG - can be loaded into the client's equipment (e.g. phone), allowing a local technician to follow the procedure and localise the cause of the problem.