The issue here is that the characters around the $t are considered whitespace from parser we use. $t will work, but if it is surrounded by ignored characters then you will get what you consider junk. The streaming API will work better for just following a single topic, but the specific query here for $t was created for stock search years ago. We likely need to think about this case and and maybe improve the parser.
Jonathan @jreichhold On Sun, Jun 27, 2010 at 8:21 PM, John Kalucki <j...@twitter.com> wrote: > You may need to do some post-processing on your end to get exactly what you > want. You could search for t, then discard all tweets that don't contain $t. > > But, it sounds like you are doing a repetitive automated search. You should > be using the Streaming API for this. Try using track for $t, it might work. > If not, track for t, then post-process. If you use search, you will miss > some tweets, due to relevance filtering. > > -John Kalucki > http://twitter.com/jkalucki > Infrastructure, Twitter Inc. > > > > On Sun, Jun 27, 2010 at 3:09 PM, Zhami <stu...@yellowhelium.com> wrote: > >> I need to search Twitter for tweets containing the phrase "$T" (T >> being the trading symbol for AT&T). When I perform a search for this, >> I get back oodles of results where "$t" is not a complete phrase, but >> the characters in some larger phrase, which isn't what I want. Alas, >> $t is quite common in 1337 (leet) phraseology. >> >> The Web interface for search.twitter.com "advanced" offers an entry >> for "This exact phrase" but I can't get that to work. I have tried >> specifying spaces on either side of the $t text, as in: >> http://search.twitter.com/search?phrase=%20%24t%20 >> But the search engine ignores the spaces and returns non-"exact" >> phrase matches. >> >> Is there any way to inform the search engine that I want exact-exact >> matches, or whole-word matches? >> > >
