I will discuss where and when matching is necessary for
OpenKnowledge peers using LCC interaction models and suggest methods
for dealing with these matching needs. In many situations, it
will not
be possible to find exact matches and in such cases we need to find
approximate matches. We then need some method of evaluating how
good
these approximations are so that we can select the best match in each
situation and then determine whether this exceeds some threshold 'good
enough' value. Much of this work is still rather speculative and
any
feedback would be much appreciated.