European Semantic Web Conference took place this year (2022) in Crete, Greece with a lot of sun, awesome people & talks, but how was it there exactly? I had a pretty booked schedule, I presented in various places of the conference from workshops to posters. If you attended, I hope you enjoyed my talks and if you’re interested in my work, feel free to contact me to collaborate or discuss (contact details)! Here’s my report about the ESWC experience as my first in-person conference from my point of view:
Day 0: Travelling & schedule
Due to the pandemic, it was ages ago I travelled, especially by plane. Weeks before the conference, I prepared all my talks, arranged a place to stay, take care of the flights, etc. Took a while to get everything in-place, but I think it will go faster next time. I will be travelling soon again in a few weeks, anyone who already knows where? Let me know on Twitter! Anyway, I hopped on to the plane without much issues and arrived in Crete without much delay in the late evening. Shared a taxi with some colleagues to the hotel. However, the receptionist couldn’t find our reservation for the rooms. After some short panic it seems it was registered under a different name, so luckily, in the end everything was resolved, I had a place to sleep for the night!
ESWC 2022 consisted of 2 days Tutorials & Workshops and 3 days of main conference. Let’s go over them!
Day 1: Tutorials & Workshops I: KGC tutorial by COST-Action
The first day, I attended the Knowledge Graph Construction Tutorial since I was also a presenter there. This amazing tutorial was organized by DKG Cost Action. We had a full house for this tutorial and even had to move to a bigger room to fit everyone in! A lot of tools and approaches where presented such as Mapeathor, RocketRML, SDM-RDFizer, Dragoman, etc. It was a great overview of all the different approaches and tools used in knowledge graph construction, including some hands-on experience!
I presented 2 tools during the tutorial: YARRRML in the morning and RMLStreamer in the afternoon. YARRRML is a human-friendly representation of RML which allows you to write a few YAML-based mappings to transform all kinds of data (CSV, XML, JSON, relational databases, etc.) into RDF using any RML compliant engine such as the RMLMapper or RMLStreamer. Afterwards, I presented the RMLStreamer and how you can use it to transform huge amounts of data into RDF. The RMLStreamer leverages the Big Data platform Apache Flink to scale the RDF generation across CPUs (vertically), but also across machines (horizontally).
I was amazed by the response of the community when using all the tools during the tutorial. I never expected that so many people would be interested in the work that I and the others of the knowledge graph construction community are doing!
Day 2: Tutorials & Workshops II: PhD symposium & KGC workshop
During the second day, I had to present my plan for my PhD (“Balancing RDF generation from heterogeneous data sources”) at the PhD Symposium track of ESWC. This track allows PhD students to get feedback from seniors in the community related to their plans, approach, evaluation, etc. I was the first presenter of the track, so I didn’t really know what to expect. The whole room was pretty filled with people queuing outside the room, but I didn’t really have stress. The Semantic Web community is constructive and nice so I had nothing to fear! After presenting my work, I got some constructive feedback and comments which I can definitely use in the next stage of my PhD. I would like to thank all the attendees for all their feedback to push my PhD to the next level.
As mentioned before, my schedule was really full which made it hard to attend other talks as I had to present in various places. While, I had to attend the whole PhD Symposium, I simply couldn’t as it overlapped with the Third International Workshop on Knowledge Graph Construction where I had to present another short paper “Continuous generation of versioned collections’ members with RML and LDES”. In this paper, we set the first steps towards incorporating data updates duing knowledge graph construction with declarative mapping languages such as RML. By leveraging these data updates, we could reduce the generation time and materialized RDF triples significant. However, these results are still preliminary, we consider to execute a more detailed performance evaluation later on.
Day 3: Main conference I: Posters!
Day 3 was the first main conference day, the day when I had to present my poster. It was my last presentation of the conference, so I had to be awesome once more in the evening. During the day various talks were given and an awesome minutes madness. I presented my poster during the evening reception where I met so many awesome people with constructive and good feedback. Thanks a lot of everyone for coming by! It was so busy I didn’t have a chance to grab something eat or drink, luckily my supervisor brought me some food for which I’m very grateful.
Day 4: Main conference II: Gala dinner
The second day of the main conference, was kicked off by Axel Ngonga with a keynote in which he explains how important representation is when doing structured machine learning. Later on, we had a session where papers were presented around dealing with multiple sources. After lunch, I went to the Industry Track to see what Siemens, Bosch, etc. were doing with Semantic Web technologies. Siemens is using W3C Web of Things (WoT) to describe complete buildings for automation purposes. It was really nice to see these W3C WoT specifications being used in the industry. After the coffee break I went to Linked Data Analysis track where a research paper around XSD binary floating point datatypes was presented. It really reminded me of a course I had a few years ago where floating point errors were discussed. The same errors also occur with XSD floating point data types used in RDF. It is worth to investigate if these errors also occur when generating RDF from heterogeneous data sources such as with RML. Even better: how to prevent or reduce these errors during knowledge graph construction! Not only the datatype may cause these errors, but maybe also during the transformation from heterogeneous data sources to RDF. Something worth to consider since they forget about it when launching the Ariane 5 rocket. A floating pointing error, caused the rocket to crash during a test flight. In the evening, we had the gala dinner with all the people of the conference, it was really nice and awesome to see everyone in a great dance mood!
Day 5: Main conference III: All things come to an end
Before you know it, it is already the last day of the conference! The opening keynote by Tova Milo was amazing and really well explained. After the keynote I went to the Software Development track, I have an engineer background after all :). One of the papers was about lazy loading of RDF in applications. I thought of that as well in the past and I can say that it really makes a difference, especially when dealing with a lot of RDF! For the last session, I picked the Reasoning Systems one because I really wanted to see a reasoner on an Arduino. They actually did it which it pretty awesome! Most of the time we as Semantic Web researchers deal with enormous amount of data and systems. This is the first time I saw such technologies being used in something like an Arduino with limited RAM and CPU resources.
ESWC 2022, my first in-person conference, was AWESOME! The bar is now pretty high for the next conferences :) I met so many new people and really enjoyed the beautifull view of Crete. I hope to see everyone soon again on the next conference or the next ESWC!
If you want to checkout my papers and posters published at ESWC 2022, you can find them on my publication page.