Software need not work exactly the same way no matter who uses it and when.
For example, the English language has perhaps 2 million words,
but most of us use only about 70,000 or so,
and of those 70,000, perhaps 60,000 are rarely used. The point is that the
10,000 most frequent words that a particular person uses might be a strong
indicator of what kinds of topics might interest that person.
Thus, those words (or their equivalent) should be used to bias
search results. It isn't relevant that a search returned 100,000 hits within
the user's database, what's important is which of those 100,000 are most
likely what the user had in mind when initiating the search.
|Context For Free|
|Star Trek Search|