Improvements

The sign-to-unicode mapping is more complete.

Apart from some white space, the data has not changed otherwise.

At the same time, the data has been generated anew with code that also generates the Old Assyrian data.

I have diffed all feature files with those of the previous version.

Fix in numerals

Some features represented numerals like 5(disz) as 5(5(disz)).
This error stems from a bug in the conversion.
It has been fixed and re-run.

New data version

The source data has received corrections.
The unicode mapping got a bit better.
The similarities have been recalculated.

More unicode mappings

The unicode mapping is improving after a list by Martijn Kokken

Metadata + unicode mapping

Better unicode mapping;
more meta data

Added parallels data

The feature sim is an edge feature between similar lines.
The value of sim for a pair of lines is their similarity, as a percentage.

Minor change

The -aftere- feature has been changed and renamed to -afterr-.
This affects some text formats in that it puts a separator character between adjacent signs in order to prevent ambiguity.

Presentable data

Data conversion has been checked.
Various features have been added to help formatting and styling the presentation of text.

Integer valued features

The features ln and col for column and line numbers are now integer valued.
The numbers do not contain primes.
If primes are present, it is indicated in the features primeln and primecol.

But I had to do something about comment lines, which got numbers like $a and $b.
These numbers now go into a separate feature lnc (character line numbers or comment line numbers).

Note that the full number, lnno incorporates the column number and the primes and the comment numbers.
This is used for section headings for lines.

More features

This version of the TF data has better features to support various text formats.