Hmmm. Very clever solution to use Google and I agree that unique people MENTIONING a term is more valuable then the actual mention.
It's wild to me that the Twitter API has not started to incorporate more stat related metrics. Seems crazy to me that a server as useful as Twitter that has become a mainstream media tool does not yet provide metrics that would be "business" useful. Hopefully soon I suppose... On Apr 29, 1:27 pm, Nick Arnett <[email protected]> wrote: > On Wed, Apr 29, 2009 at 8:57 AM, JoshL <[email protected]> wrote: > > > Does anyone have a good suggestion for how to obtain the data needed > > to know how many mentions of a specific term occured PER day over a > > given time period, such as two years? Omniture's SiteCatalyst seems > > to be doing it somehow. > > Can't be done right now, since there is nowhere near that much history in > Twitter's search index. Even when there is, there will undoubtedly be an > upper limit on results, which will prevent you from getting all the history > for popular terms. > > However, if the data were available, the methodology would be fairly simple. > You'd search on the terms and then iterate through the search results, > counting unique mentions by day. I'll suggest that for most purposes, the > number of unique people mentioning a term is more interesting than the > number of mentions (I've done a lot of that kind of analysis). > > I guess there's another possibility - use Google, which has more history. > > e.g.http://www.google.com/search?hl=en&rlz=1C1CHMI_enUS291US307&q=site:tw... > > For that term, there are about 56,000 results... but you can't get more than > 10,000 results from Google. And you'd have to either parse the resulting > pages to extract the status messages or just capture the screen names and > use the Twitter APIs to get the statuses... fairly horrendous amount of work > to get the data. > > For the sake of completeness, I'll note that you can get beyond 10,000 > results from Google by excluding terms, but there are also daily limits to > Google API queries. > > And... knowing that Twitter Trends is doing essentially the same thing over > the short term, I would suspect that if there's a need for this, Twitter > will eventually tackle it. > > Those who are doing it now must have captured the data earlier if they have > a year's worth. > > Nick
