alexjoz / mathtextrecognizer Goto Github PK

We doesn't know what the output might be, as it can be defined by the user.

Also, the symbol usage should be increased in all the databases.

Many of then work wrong.

A dialog should be displayed to the user, so he can choose to replace the
older symbol with the just processed one ("replace"), to ignore the newer
and keep the older ("ignore") or append the new result ("add").

This will involve changes in the GUI, as in the database format.

Character information databases should handle the cases in which multiple
symbols image are associated with the same pattern. 

In the current implementation, a conflict appears, so if there is a symbol
matching a pattern, newer symbols associated to the same pattern can't be
learned.

When a conflict is detected, instead of just warning the user and
discarding the newer symbol, a set of options should be given to the user:
discard the new symbol, add it to the pattern's set, remove all the symbols
associated to the pattern and add the newer symbol instead.

Nowdays, formula's pattern matching is done in a way that makes imposible
teaching the program new formula patterns.

A pattern editor should be added, which would create the structure needed
to make the pattern matching variable.

 The dialog doesn't get launch.

Also, the user should be able to define a custom set as the default,
overriding this.

When in educational mode.

If we don't know the symbol, allow the user to learn it. The refactoring of
the learning interface should be made to make this possible.

The dialog allows an empty text without complaining, and the validator
triggered when the goToTokenizer button is clicked doesn't detect it either.

In particular, it would be very useful to store usage info of the symbol to
provide a statistical background so it's easier to provide the user a
choice between conflicting symbols.

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?


Please use labels and text to provide additional information.

If a rule referenced in a rule call doesn't exist, the syntactical analysis
process fail miserabily.

It is easy to mistype a rule name so those errors are common. Therefore, a
validator should be implemented, posibly before closing the syntactical
rule manager.