My 7 key takeaways from Semantics and LTI conference(s)

On September 14 and 15, 2022, I attended the joint conference of Semantics and LTI. It felt like going to a wedding and only knowing one side. You essentially started a conversation by asking „Are you on the bride's side or the groom's side?". However, the overlap of what both parties are doing turned out to be immense. Language geeks talked about taxonomies and intelligent data a lot, while the semantics crowd always mentioned terminologies, also. Shows us how spot-on we are at Kaleidoscope to integrate the worlds of Semantic AI, taxonomies, knowledge graphs, etc. into our Kalcium Quickterm terminology platform.

Here are my 7 key takeaways from Semantics and LTI:

1. „No-brainer ideas get killed by the lack of data." Or put differently: „Data will make or break the future." That was the summary of the keynote by the owner of the IKEA knowledge graph, Adam Keresztes.
2. On language technology insight, it was interesting to see the results of the Nimdzi research. Decisions for a new TMS are taken 47% based on the UI, 37% on Integrations, and only 16% on pricing. MT capabilities are surprisingly low at 11%. Nimdzi sees upcoming trends in string localization and connectors. And on the MT side, the challenges are speed, quality, and monitoring of output.
3. From an LSP perspective, the presentation by Kirti Vashee was intriguing. He argues: As customer experience is becoming ever more important, companies need to be multilingual in real-time and in all their communication. This changes the production model from „broadcasting" to „sharing", which I will come back to later.
4. Much discussed topics on the Semantics side were data fabrics and graphs. The advantage of graphs in AI is that they allow you to follow the decision process and influence / fix it. This leads us to "explainable AI" and away from the black box.
5. A great quote by Andreas Blumauer from Semantic Web Company is worth mentioning: „When you talk about interoperability, you need to realize that part of interoperability is language." Again: Semantics people are talking about language, just like language people are talking about Semantics, but the take is different.
6. A remarkably interesting „take-away" from fyrfeed, a content creation platform utilizing AI abilities, is that the mention of AI in content creation services scares away customers. This stands to show that translation, as opposed to original content creation, is still a very commoditized thing. Why do customers shed away from AI in content production but are eager to use AI (=NMT) for translation? A lot of work still in front of us!
7. Finally, to wrap up the staggering number of 109 presentations in two days, Jochen Hummel, Britta Aagaard, Pedro Vasco, and Bruno Herrmann introduced the concept of "LangOps". First mentioned in a Forbes article, #LangOps draws a parallel to #DevOps, arguing that Language should be a cross-company, infrastructural, cross-team approach to run any business process in multiple languages. While it goes against my credo of not talking about language anymore, the concept is interesting, because it addresses the technology buyers directly and it calls for a paradigm shift. „As long as we are only trying to make translation better and faster, we won´t get anywhere". (Pedro Vasco).

The final highlight of Semantics 2022, of course, was the prize we got for our panel „Towards One-stop-shop Language Intelligence Partner Spheres in the Cloud", together with Erik De Vrieze from RWS, and Raniero Romagnoli and Vincenzo Sciacca from Almawave, chaired by Josef Kubovsky from Nimdzi. LTI awarded us the prize for „Best Language Intelligence 'Maturing Technology' Presentation". Wow. Thank you, Philippe Wacker, and I am sure looking forward to the next event of this kind!

These are my 7 key takeaways from the joint Semantics and LTI conference(s). Now, I'm interested in yours!