---- Stop List ----
the
of
and
in
to
a
is
for
as
by
that
are
with
on
from
be
or
was
it
an
at
which
this
has
can
have
also
used
such
more
its
not
than
may
into
but
most
some
they
there
will


---- Note ----
This is just a small sample to illustrate a point. A full stop list would be much larger.

---- Choosing cut off values ----
We choose the cut off values for our stop list based on frequently used terms that had little or nothing to the context of the query.
Further, removing them would cause little to no meaning to be lost.
In the explanation section is an example.

---- Explanation ----
Looking at the Unigram data, I see a ton of characters that we could have ignored.
We could easily ignore all hashtags and # signs as they do not add anything to the results.
I think we could also safely strip "$" from numbers and mathematical notations like: kw·h
This would grately reduce the size of the unigram list, consider the following example:

500 -- 218
500% -- 3
500+ -- 1

Had we stripped mathematical notations, these 3 records would just be one, grately reducing the length.

Now, when dealing with common words, we could easily remove several of them:

"the", "of" are examples of the top 3 words, we could easily remove them without losing the context. For example:

The Spanish Coast -> Spanish Coast
The Coast of Spain -> Coast Spain
The meaning is retained, yet the index size is grately reduced.
Of course, anything we do to reduce the size of the index, we must also do to the query.
If we remove these terms from the index, we must also remove them from the query.