Re: The wildcard-filling problem

> The main problem with this solution is how to get to
> the object from the original sentence string. Keeping
> the mapping list in sync with the normalization
> process can be a real pain, not to mention
> performance-costly; besides, I have found no way to
> correctly update the list when a substitution adds
> more than one space to the sentence (on the up side, I
> could not think on a good example of this, so it must
> be a pretty rare occurrence). Finally, the creation
> rules are promptly defeated by space groups on the
> input string, thus creating the need of a
> pre-processing step -- and special care must be taken
> not to add extra spaces with each substitution.
>
> Despite these lower-level problems, I sill think that
> on a higher abstract level my solution is relatively
> simple and ellegant, and that in time I will be able
> to iron out any implementation shortcomings. But of
> course, if anyone knows of a simpler and/or more
> efficient alternative, I would be glad to know.