I have actually thought about this before! I use the Wendler 5/3/1 workout plan myself, and there are plenty of apps for it, but maybe building a KNIME data app specifically for it would be a fun “exercise”.
I’ll add this here, as these were ideas that came up in another thread, though it is worth capturing the ideas here. Apologies if these are previous challenges, I don’t follow them so don’t know what you might have done previously.
Adding Metadata To Articles
The original thread was asking how to match keywords to text in an article so that codes associated with the keyword could be appended to the articles metadata. The proposed solution used the Text Processing nodes to clean-up the articles and keywords so that they could be matched. However, the matching process is imperfect and a number of documents remain uncoded.
Example documents would be job adverts posted online, where codes need to be added to identify the industry segment of the employer and the type of job on offer; or customer posting notices of requirements for the supply or goods and services, in which case adding codes to identify the industry segment and products would be useful.
Whilst the solution used the Text Processing nodes, this requires manually created keywords and coding lists, which need further manual intervention as the keyword matching is exact. However, with a sprinkle of KNIME magic it should be possible to look at the articles that have been coded and create a model which predicts coding for those without direct keyword matches.
Autocorrecting Manually Entered Text
The second challenge is to correct comm typographical mistakes in manually entered text. This is quite a common task and it would be nice to have KNIME tools and workflows to clean up text prior to further processing. Again, this task could be done with a manually created database of keyword matches for common mistakes, however, it can’t capture everything and is hard to maintain. KNIME is magic and I am sure that with the right application of brain power, creativity and teamwork there is a solution to this problem.
I would like to see different ways of handling flaky APIs. The GET and post nodes are great, and I use them all the time, but you get so many different strange errors back from APIs that even the best designed workflow will fail due to external APIs. Would just like to see how the hive mind handles these scenarios.