building a upweight and downweight list of search terms -when words are buried in a model attribute

[harijay]

Hi I have a small database of few thousand entries video files of
screencasts ( video screen-capture ) of software.

Each video has a title which is a sentence and a detailed description,
that can be a few text paragraphs. I am having haystack index both of
these fields in my model.

I am using the solr backend and the results are really quick and
accurate, for the most part.

However I want to have a list of words to upweight and downweight in
the results. Since these are all tutorial videos I want to downweight
terms like "tutorial" "demo" "howto" etc , similarly I want to up-
weight terms that are outside the dictionary and could be names of
software that these videos are demos of...for eg Chempaint ,
Jspecview , Excel ( interesting because it is also an english word"),
powerpoint etc...

How do I build such a whitelist . The models sadly dont have a "name
of software" as a model attribute , so I am trying to introduce this
as a boost term to my haystack search. I also find that haystack
returns results even for just words like "to" , "why" etc..How do I
not return any results for common words..or downweight them into
oblivion.

Any ideas on how to achive these. Sorry I am new to rolling my own
search.

Thanks
Hari