Just need a little clarification
Ok... what do you mean by filtering out packets with "unwanted words or phrases" ... what specific words or phrases are you trying to filter out (Example: Like profane or sexual oriented content on web pages?)
Just trying to understand what you want to filter out?
Sounds Like you are complicating this
I think a better question for you to ask it what other routes you can take to filter content. Trying to filter at the packet level may not be what you need.
For email their are different filters out there depending on what type of email you use, ie outlook, eudora, etc..
Then their is software that regulates what people on the net can surf.
Trying to filter at the front end would take too much time and their is already software out there that is free to use on the client end.