After being extremely impressed by using OntoWiki for Geoknow, we setup OntoWiki for the LinkedSpending project. However due to it having other properties, see the LinkedSpending paper, there were several problems. Also I think there were some forbidden paths like /resource, so if you want to put your instance for example under http://linkedspending.aksw.org/resource/example, it would break. But was many years ago, it may not be like this anymore.

Large amounts of data

OntoWiki didn't seem optimized for such large amounts of data, even with some modifications made by Michael Martin to simplify some OntoWiki index queries.
Some of the SPARQL queries generated by OntoWiki would overload the SPARQL endpoint and cause drastic slowdown or even crash of the website.
This could also be due to the resource restrictions of the virtual server, which has 6 GB of RAM and 3 CPU cores (may have been different at the time).

CubeViz plugin

While the CubeViz plugin was one of the few, or even as far as I know the only possibility at that time to visualize RDF data cubes in an editor, it also seemed to be hindered by the large amount of data. Also, the sparseness of the datacubes made it very hard to get meaningful visualizations. The paper contains one but it took a large amount of trial and error to set the facets to the correct values that results in a meaningful diagram.

Complex (meta-)structure of the RDF Data Cube Vocabulary

RDF Data Cubes are modelled using the RDF Data Cube Vocabulary and RDF is only a lower layer here. However OntoWiki works on the RDF-level so of course it cannot verify that your data cubes are correctly modelled or prevent you from breaking them. This is a general problem with OntoWiki and all other RDF editors, I don't know if there is even something you can do here. Theoretically, OntoWiki could read in some ontology or meta ontology or SHACL shapes or a similar mechanism and verify and protect your data with that in mind.

Break on Ubuntu 20.04 LTS Upgrade

An upgrade of the server from Ubuntu 16.04 to 18.04 went without problems but when going to 20.04 after that something broke.
I don't remember what we did exactly (maybe something with the PHP version?) but I think Nathanael helped me to fix it after that.
Docker is probably the solution to that but I wasn't comfortable with that so we fixed it in a direct installation.

The semantic web is to my knowledge the most suitable tool to model any type of information. Currently there is no front-end software available that enables a lay audience (99% of the world's population) to make use of this quality. Wiki software is promising for personal and collaborative knowledge management but current wiki-software lacks in information modelling through linked-data, therefore missing the potential that linked-data systems have over simple text documents.
The vastness of representation possibilities (front-end) and their connection to the underlying data (back-end) represents the biggest challenge. I have not been able to use OntoWiki, but having familiarised with the concept and having seen some screenshots, it appears to me that it makes a great step forward to achieve this goal.
Hopefully there will be a functional implementation again, or a good alternative.
Note: Another very promising ecosystem that is a great step forward in this regard is Solid: https://solidproject.org/, which is a semantic back-end open for front-end contributions. It would be fabulous if the new OntoWiki were a front-end for it!

I wonder if OntoWiki could be transformed into a set of MediaWiki Extensions which cooperate with Semantic MediaWiki and Wikibase.

GeoKnow

I first used OntoWiki around 2013 for the GeoKnow research project to let the project team collaboratively edit the research project structure (work packages, tasks and deliverables) of the GeoKnow project and to publish it on the web page.
It was modeled around the FP7 research ontology http://purl.org/research-fp# created just for that but it doesn't seem to be used by anyone else anymore and the link is dead as well.
My PHP skills don't extend beyond small websites and I didn't know quite how to set it up but I think Michael Martin did it for me on the GeoKnow Ubuntu virtual server. One early pain point was the default setting to give everyone write access, which makes sense for a Wiki but caused automatic web crawlers to traverse delete links, causing many mysteriously deleted resources until we figured out that there were no mysterious bad guys and disabled anonymous write access.
This issue persists to 2021 (using OntoWiki 1.0.0), so my recommendation is to disable write access for the anonymous user by default or at least prevent deleting resources by opening URIs alone. This problem is described in more detail in https://symbolicdata.github.io/OntoWiki.

After that it worked really well and I think it was a perfect fit with OntoWiki and the site extension because there are just a few classes like milestone, project, deliverable and so on, and those could are shown both in numbered lists as well as in detail views that are hierarchically linked using OntoWiki site-extension templates.
For example, http://geoknow.eu/wp2.html shows the detail view of work package 2 but also shows the other work packages in the sidebar and all tasks associated with that work package in a list below.
I was really impressed by how we could create and edit such a nice website for a multi million euro research project with such ease, including multi-user concurrent access, backups and so on.
I think if we went to a web developer or company and asked for a website with that feature set including a CRUD editor, backups and so on it would have been quite costly.
Finally, when the project was finished and there were no more changes, we wanted to keep it online indefinitely but also minimize the effort to maintain it, so the we just made a static copy of the entire website using the Linux wget tool with recursive traversal of the generated HTML files.
That static web site is still rock solid many years later and will probably be for as long as the server exists, because the result is just static HTML without much JavaScript or other dynamic content that could somehow break.

Properties

• small amount of data
• site extension
• simple structure

All in all, full marks from me for OntoWiki + site extension for the GeoKnow use case.

SNIK

• https://www.snik.eu/

• SNIK is a large ontology with the SNIK meta model

• Challenges caused by meta model and OWL restrictions: one fact is represented by several triples, OntoWiki does not warn user when deleting parts of such a fact, rendering it invalid. This is explained in more detail in #2.

• OntoWiki is used for selective changes and Semantic Web inexperienced users

• SPARQL Update queries are used by technical users and for sweeping changes involving many different resources at once.

• This presents a problem with the history function of the OntoWiki, as it isn't the only place where the data changes.

• Backlinks would be useful for that project, OntoWiki shows all ?x ?y ?z triples for a given subject ?x but it would also be great to see the inverse ?z ?y ?x, as sometimes the direction is arbitrary, especially with symmetric properties where we don't materialize all implied triples.

• We linked to OntoWiki from SNIK Graph and the SNIK LodLive RDF Browser. This is possible because you can specifiy the graph and model in the URI. It would be great if there were an automatic graph determine function though, where you could just give the URI of the resource alone.

• The "bot attack" problem of accidental resource deleting when the system configuration gets reset still (version 1.0.0) occurs. See https://symbolicdata.github.io/OntoWiki.