My view of the desktop search hackfest

I was attending the Desktop search hackfest in Berlin last week. It was sponsored by Nokia which did a very good job in choosing the location and organizing everything. Big thanks to Nokia for all the work and money they put in. Unfortunately, it was more of a discuss-fest instead of a hackfest. Originally, my plan was to write Python and C# bindings for xesam-glib and plugins for Deskbar and Gnome-Do. It become clear quickly that this ambitious plan doesn’t work out. The schedule was packed with discussion about Xesam Ontology, Search API and Query language. Basically, the discussions about this three topics protracted over the whole hackfest. Mikkel already summed up the conclusions nicely on a wiki page.

In addition, it was agreed on to write a Xesam test suite. It’s purpose is to check how good search engines support the Xesam spec and also provide some numbers on the performance. Once that’s ready it should be really interesting to see how the search engines perform and the improvements they come to their minds to get the number one spot. Personally, this was the most interesting conclusion from the hackfest from my point of view. Although, I managed to write Python bindings for xesam-glib 0.5.0, I don’t know what they worth in the end. At the end of the hackfest the Xesam search API had not much to do with the API before. This means that large parts of xesam-glib must be re-written to incorporate all the changes. Actually, I think the big changes mainly in the search API will result in a setback first, because it will take a long time until search engines support the new search API and query language. And when there are no search engines that support this spec there won’t be any applications either. At least I don’t want to write a module for Deskbar for a spec that isn’t implemented. Maybe this changes are worth it in the end. Time will tell. At least once the test suite is implemented it should make the life search engine developers a lot faster and speed up development. Nevertheless, I had fun in Berlin and it was great meeting some of the people involved this field and listening to the ideas they have. Now one has to watch and wait.

Avatar
Sebastian Pölsterl
AI Researcher

My research interests include machine learning for time-to-event analysis, causal inference and biomedical applications.