insufficient documentation

[jairosg]

I have just started using this tool for my master's thesis and I have noticed that some documentation could be more complete with more examples in tasks like CrossEntropy filter and similar ones, since you can't know what to put in some parameters if you are a beginner and it leads you to get too many errors that you don't know how to fix.

[svirpioj]

I do agree. Unfortunately writing documentation is time-consuming, and time is always limited. Some quick hints:

• Start with the default parameters. Use the opusfilter-test script to see how a single filter works on (a part of) your corpus. Especially use --removed to store the pairs filtered out.
• For setting up something more complicated like the CrossEntropy filter, take a look at the example configurations in the repository. Our idea is also to collect more example configurations to OpusFilter-hub (although at the moment it's quite empty).
• The separate OpusCleaner project develops an interface for selecting filters and suitable parameters for them.