Hello! I’m trying to use WebLicht service to obtain parts of speech from a tokenized word tier. The process doesn’t work because of the occurred error 400 or 500. I don’t understand the problem: I’m using a child tier, its stereotype is symbolic subdivision (as ELAN manual says). What is the error?
Hello,
In the WebLicht scenario ELAN uploads data (text) to a web service and, on success, converts the results to annotations. The errors 400, 401 and 500 indicate errors on the server side, for the exact phrasing you can have a look at this Wikipedia page.
I just tried several services and receive the same messages for many of the services, some still work. Although in principle the problem could be a temporary problem per server, but it is more likely that the way ELAN accesses the services is not supported anymore. It’s probably already several years ago that using the WebLicht services and tool chains in a browser requires the user to log in. And since ELAN accesses the services without requiring log in information from the user, this might be the reason behind the errors.
So we probably have to update the way ELAN accesses WebLicht if we want to maintain this functionality. I believe the preferred way then would be that the user first creates and downloads a tool chain in the WebLicht interface and that ELAN sends the chain information with the input text to WebLicht.
So, I’m afraid there’s no short term solution to this problem. If you already tried a particular service multiple times and the errors are consistent, it is likely that that service won’t work anymore from within ELAN.
-Han
Thank you very much for the clear explanation! I will try to do as you suggest, i.e. create and download tool chains in the WebLicht interface. That way it should work. Thanks again!
Just to be sure: using the tool chain from ELAN still has to be implemented, it is not working now. Running the tool chain on your data in the web interface should work (and maybe storing it, but time-aligned integration of the data in ELAN won’t be too easy.)
Thanks for the advice keep supporting