From manual writing to automatic generation: an insight into Asylos research

by Marius Jammes

When writing a Country of Origin Information (COI) report, researchers run the risk of spending much of their time on writing and formatting text rather than on research itself. As Asylos researchers are volunteers, it is especially important that they are able to focus on the latter. Furthermore, Asylos’ network-like structure and emphasis on pooling skills and knowledge encourages the use of a system that allows researchers to work collaboratively.

In order to reduce time spent on formatting and to set up a system that facilitates collaborative work, we evaluated a number of tools for collaborative writing, like Google Docs or Etherpad. We also considered to use markup languages to let researchers focus exclusively on the writing process and let the computer do the formatting (e.g. LaTeX, Markdown). However, even when using templates, those tools and languages do not really fit Asylos’ current needs. While tools for collaborative writing make collaboration easy, they do not solve the formatting issue. The problem with markup languages is that, as far as we know, there is no turnkey and user friendly solution that would make collaboration and conditional formatting easy without much coding.

Country of origin (COI) reports are structured texts, based on quotations and with only little space for researcher’s own phrasing. This means that it is theoretically possible to automate almost entirely the writing and formatting process, using conditional formatting and raw data. Researchers would only have to save their findings in a spreadsheet or a database and click on a button to automatically generate a report. What we were looking for was a tool that would allow us to implement this without spending too many resources on coding.

To do this, we decided to use Google Docs together with Ultradox. Ultradox is an online tool facilitating automation of some tasks. With a bit of work upstream, it enables researchers to produce their reports in a few steps:

1. They fill out a form each time they find a source that addresses their research question (answers to the form are stored in a spreadsheet);
2. They load the spreadsheet into Ultradox;
3. They click on a button to generate the report.

Of course, this last step requires some work upstream as it is necessary to code the templates that Ultradox is using to retrieve data from the spreadsheet, print this data in the report and format it in the correct way. The infographic below summarises the global process and its logic.