This is a guest post written by Naveen Thattil, CEO and Founder of TripTwit.com.
Real-time and location based searching have been trendy topics of discussion for the last year or so. Google and Twitter both offer the ability to do real-time searches for a specific geographical location with data coming whole or in part from Twitter. While the potential is great, there are still some challenges ahead.
As it stands now, the services are built to search for keywords appearing in tweets originating from the specified geographical location. While this has the potential to deliver relevant results, it also returns noise. Perform a search for “happy hour” in Chicago and you’ll get mixed results.
You will see data about happy hours happening in your area right now and you’ll see people tweeting about hurting from “happy hour” last night. Presumably the former is more relevant than the latter, but they are viewed as having equivalent authority and simply returned in a time ordered stream.
Lack of authority also results in real-time results getting spammed, which has been well chronicled over the past year.
2) Geographical Scope
As mentioned above, these services grab tweets from a specified geographic location, but does geography always correlate to relevancy? Let’s say I’m searching for “SeaWorld” in Orlando and I get results from the Orlando area. The results may provide decent data, but am I seeing all real-time data that is relevant?
A twitter user named CompleteOrlando is a perfect example of this dilemma. Most of their tweets have to do with deals in the Orlando area, including Sea World, yet they’re based in London. Although their tweets are highly relevant to Orlando, they will never show up in Orlando based searches.
Location alone isn’t the silver bullet for relevance.
3) Data Availability
The only way you can truly tell where a tweet originates from is if the user has geotagging enabled. By default this feature is disabled. One study estimated geotagging usage at 0.23%. That means there are millions of relevant tweets that go “under the radar” that will never be returned in user searches because the search service cannot tell where the tweet is coming from. The results you see are really the tip of the iceberg.
If you’re simply searching for what people are saying in Los Angeles about “justin bieber” then the results you get may be just fine. If you’re looking for other types of information like what is going on, what is there to do in a specific area, etc. then these services have some fine tuning to do.