P.R. Pottelberg suggested to solve the problem of homonyms by numbering of words on the web.
2.Refine your query
TopDollar commented to the above suggestion and said that
If you want to search for "light" as in to light a fire, just do a search for "light fire.
IMHO TopDollar doesn't accept the fact that there is a problem. There are many search experts that know how to search and still fall into that pit. A man can not control all these hidden homonyms – but a computer can.
P.R. Pottelberg's solution will be very clumsy since there are words that have multiple meanings and users will not like scanning long lists only to get a better query (like the long list of synonyms in surfwax)
The complexity of the problem grows when you think about homonym phrases (not only single words) like "looking for love" .
Directories have an advantage over search engines in this case since they put each term ("light" in this case) in a separate category.
6.Search is so fast
I don't know a search engine that tries to solve this problem. Maybe there are such search engines but anyhow it doesn't seem to me to bother users too much. It looks as if every body is happy with the current situation. Search is so fast that you can refine it in a blink of an eye.
7.Human answering services
Human answering services like Wondir and Yahoo Answers don't seem to have this problem. People understand queries by context.
With Qtsaver it is easier for users to understand if they fell to the homonym pit because of the proximity of the short chunks. They can see immediately if the term is used wrongly.
9.Micro content tagging
I believe that micro content tagging can help solve this problem. Those who fall into the homonym pit can tag the micro content that had this problem and warn the future readers.