Leif-Jöran Olsson is a language technologist and XML enthusiast with a long history in the Swedish solidarity movement. I sat down for an interview with Leif-Jöran and asked him about his background, his education and the various projects he’s been involved in.
Stian Rødven Eide: A major part of your work has centered around language technology (LT). What was your point of entry to this field, and how does it relate to your dedication to Free Software? Were you already interested in Free Software when you started your education?
Leif-Jöran Olsson: I was initially very interested in usable design, and joined the Mechanical Engineering programme. But, after two years, I found out that human communication was much more fascinating. My introduction to Free Software came after Gymnasium, where we mostly used proprietary software like turbo c++ and turbo prolog. Since I come from a rather unprivileged background, and could not afford to buy software, this prompted my search for free tools. I later moved to Uppsala with my own family and attended the Master’s programme in Language Engineering, which, coincidently with my search for education in human communication, was started in the autumn of 1994.
SRE: Some of your earlier projects have had a focus on machine translation. How does that relate to your later involvement with Språkbanken (the Language Bank) at the University of Gothenburg? Has Free Software played a part in your work?
LJO: While not the best venue for Free Software historically, machine translation was one of the primary areas when I worked in the Department of Linguistics at Uppsala University from 1998 until 2003. Here at Språkbanken, however, we have a heterogenous research environment for language technology infrastructure, primarily focusing on Free Software. Which is really great. We are not doing machine translation at all here. Being a rather shy business, machine translation has got many proprietary and secret tools involved – quite contrary to Free Software ideals. Instead, there is a focus on hard results, meaning that statistical methods, which are cheap in labour, are favoured. The machine translation research/work in Europe is mostly carried out in EC projects with large companies involved. This makes the real knowledge and gain for society rather small. In Uppsala, we did rule based transfer translation and chart parsing, which connects more to linguistic theory than statistical methods, and one of my tasks was to manage a controlled vocabulary. That makes the translation easier. But here, in my work at Språkbanken, Free Software has played a major and contributing role. I am also grateful to the director of Språkbanken for letting me use some of my time to work on eXist-db.
SRE: That leads me to the next question. One of your most active software projects at the moment is the eXist XML database. Why did you choose to get involved in that project? What are the advantages of using eXist-db rather than an SQL database?
LJO: We had been working with sgml and later xml-technologies for a long time, annotating the corpus materials used in the research. We were using eXist-db in our work and wanted to contribute back. This resulted in an active involvement in the project. SQL databases are good for strictly regular or structured (the S in SQL) relational data. Xml on the contrary is all about hierarchy and sequence. This is the power of the information model. Making irregular relations and annotations of
language material are very good examples for using xml technologies. Many people draw the conclusion that xml is too verbose and bloated, confusing the serialised human readable format with the information model. Remember, there are highly compact binary serialisations too. Almost all previous and current LT tools are using different input and output formats, which makes the interaction hard. Being so easily able to do things, like transformations of materials with standard tools, is invaluable. Since we are working on infrastructure, it is a natural choice to use an xml database, since you can avoid the overhead of parsing the data every time you want to use the linguistic annotations and corpus materials in interaction with yet another tool. We also have the sematic web technologies coming. Of course, you are better off with a relational database in a data shuffling situation, but, as soon you need to do irregular, read hierarchical and/or sequential, queries, it mostly boils down to a few easily intelligible rows in XQuery, rather than pages of SQL code.
SRE: You have also taken it upon yourself to maintain the recently liberated bookkeeping software JFSAccounting. As such software often needs to be adapted to every country’s specific laws, Free Software solutions are not always available without a certain amount of work involved. Do you find that there still are missing pieces in the Free Software ecosystem, with regards to the basic tools needed to run a business or organisation in Sweden?
LJO: Accounting software has certainly been one of the map’s white spots, and administrative tools for managing organisations are generally scarce. This I had experienced first hand through my involvement with solidarity movement organisations, and that’s why I took the opportunity to begin liberating the accounting and administrative tool JFSAccounting. The first publicly available liberated version is to be released at FSCONS 2009 (I will prepare fribokföring.se for the
promotion of this to organisations). The piece missing in the tool is a member register (matricle), as it initially was aimed at businesses. But the customer register part can hopefully be adopted with the right terminology for a release next year.
SRE: You’re also involved in the Swedish Syndicalist movement, especially through SAC (a federation of local workers’ unions). According to its principles, SAC is built upon political independence, a decentralised structure, local democracy and solidarity. To a certain extent, this seems to mirror some of the basic values common in the Free Software movement. Do you think that workers’ organisations such as SAC can help Free Software adoption among businesses and public institutions?
LJO: Actually, it was quite a hard job to make the federation accept a policy on primarily using Free Software, something that was finally achieved during the spring of 2008. The federation’s servers have been running Debian GNU/Linux for years, but it was much harder to get a policy for using Free Software on the client machines. Fear of the unknown and the comfort of the habit were the main reasons for this. There are still quite a few Free Software advocates in the different local unions, so yes, I think it can help the adoption of Free Software in other organisations. Many are engaged in several local, national or international organisations beyond their union. Eventually, people get used to the concept of Free Software and regard its freedoms in the same sense as the working class struggle. They realise the common ground they share with the Free Software hackers, and then, they don’t want to go back to proprietary software.
SRE: Through your involvement with SAC and your own company, aptly named Friprogramvarusyndikatet (The Free Software Syndicate), you have also established Serengeti, a network for solidarity and Free Software that offers free hosting for non-profit organisations, as well as a mailing list for discussions. Can you tell us a bit about the background for Serengeti? Do you have plans to expand its activities?
LJO: As I mentioned, I have met quite a bit of fear of technology, and an ignorance of the negative consequences from putting yourself completely in the hands of proprietary market actors. At the same time, many people that are attracted to Free Software are afraid of politics and only see Free Software as neutral and apolitical. This resulted us forming a loose network called Serengeti. We are aiming for a more stable network that can promote the use of Free Software in solidarity movements, and also help bridging the surplus of knowledge from therein. Building on tradition, we do it bottom up, starting out with a mailing list.
Our warm thanks to Leif-Jöran for taking the time to answer our questions. You can read more about him and his projects on his Gothenburg University page.