Future tasks for match frequency-based highlight extraction

[tsaglam]

As soon as possible (after #2563 is merged):

• Remove less-performant strategies, keeping complete matches and windowing.
• Remove superfluous weighing function.
• Check and adapt default parameters if necessary.
• Parallelize the frequency calculation¹.
• Check if FrequencyAnalysisOptions can be simplified with the record builder.
• Deduplicate FrequencyStrategy and FrequencyAnalysisStrategy by deleting the enum.
• Dissolve FrequencyUtil if possible
• Dissolve MatchFrequency
• Ensure it only runs when the flag is set (also only creates overhead/objects when flag enabled)
• ensure if it actually overrides the result (and ensure in JPlag.run the syntax makes that clear)
• Try to make Submission.getSimilarityDivisor() non-public again (if possible)
• Check if renaming bug has been resolved (isFrequencyAnalysisEnabled)
• Check if public method should be private (e.g. MatchWeightCalculator.weightAllMatches is only called from tests)

In the future:

• Ensure compatibility with asymmetric matches (currently, tokens of first submissions in a match are used)
• Instead of enabling frequency analysis per CLI flag, it could be enabled by default but fed into a separate similarity metric (this is only viable if the performance is fine for large datasets).
• Especially rare matches could be visually highlighted in the comparison view to draw attention during human inspection.

Footnotes

1. should be trivial for operations that work on a on-match or on-comparison basis. ↩