Twitter ups its search game: How it compares against G+ and Facebook's Open Graph

Twitter is on a roll this week. In addition to pumping up its Windows 8 app, the self-styled information network upgraded the search capabilities on Twitter.com.

Twitter’s new search features are supposed to make it easier to find accounts and recent photos related to your search topic, as well as view a stream of topical tweets as you could before. Twitter also added the ability to view recent searches and what it calls social context to your results, which is just a fancy way of telling you whether you’re following an account or not.

Searching Twitter

At first glance, the new search results page looks the same as it did before. You get a main section of search results and off to the left are a few options to refine your results. The search page also lists a few accounts you might like to follow that are unrelated to your search, as well as the latest trending topics on Twitter.

Twitter’s revamped search results page

But look a bit deeper and you can notice some changes. Off to the left is a new option to refine your search to view only photos, in addition to people or the main stream of results. In the main search column, you get a few popular tweets as you did before, followed by a number of accounts (called ‘People’) related to your search topic. That is followed by a section of photos, and then more top tweets.

Twitter’s new search results set-up can vary a little bit depending on what you search for, but it will be more or less the same for each query.

A nice addition to Twitter search is the ability to see whether you’re following an account as you type your query.

When searching for “Google,” for example, the drop-down menu below the search bar shows account suggestions such as @Google, @GoogleNexus, and @GoogleIO, and whether or not you’re following those accounts.

Twitter’s live search suggestions show whether you’re following related accounts.

Twitter has always been good at finding recent and popular tweets that contain your search keywords. The same goes for images since photo results are based on finding keywords in tweets.

But in my tests, account suggestions were hit or miss. A search for Google Glass brought up two Twitter accounts related to Google Glass on the results page: @googleglass and the now defunct @projectglass.

Twitter must not depend very heavily on account activity for its search results considering @projectglass currently has just one tweet from July 12, nearly a month ago. However, it’s worth noting that when you drill down into the people section of the results page, you get a much wider of Google Glass-related users.

A second search for the newly announced Moto X smartphone, however, did not bring up any relevant ‘People’ results. Twitter suggested three “related” accounts: Racer X, a motocross account; X Games, an action sports event; and SpaceX, entrepreneur Elon Musk’s rocket and spacecraft manufacturing company. Obviously, none of those accounts have anything to do with Motorola’s smartphone, but they do all have the letter “X” in them.

Twitter searches seem to rely exclusively on keywords without any server-side intelligence used to tease out meaning or relationships between search terms and relevant accounts.

To be fair, applying intelligence to search results is a hard problem that Google, the most dominant search company in the world, has only begun to crack. So perhaps it’s too early to expect Twitter to offer such an advanced feature in its search product.

Nevertheless, relevance in search results is becoming a big deal for social networks as the companies try to tease out practical uses for their services. Twitter, for example, is known as the go-to network for information on current events when the news is so fresh that it may not have even hit the major news sites yet. So, searches on Twitter should specialize in surfacing the latest tweets related to your query, with account suggestions used as a secondary factor.

Searching the graph

Facebook’s new Graph Search helps guide your searches using plain-language phrases. You could use Graph Search, for example, to try and find friends or people nearby that like the Moto X, a local restaurant, or Showtime’s hit show Dexter.

Facebook’s Graph Search helps you formulate plain language queries.

The question with Graph Search, however, is whether you really need to do these kinds of searches at all. Perhaps it would be helpful to know if any of your friends have traveled to Bogota, Columbia, if you’re looking for travel tips. And every now and again it’s handy to see a list of photos a certain friend has been tagged in or public photos of a specific place. But those kinds of searches aren’t really an everyday necessity.

Facebook’s search product also has a basic function to help surface people and pages that may be relevant. But similar to Twitter, Facebook relies heavily on keywords.

You can still do basic keyword searches on Facebook.

So, a search for “Moto X” on Facebook brings up page results for that name, but no page suggestions for Motorola or its owner, Google. A search for “iPhone” returns similar keyword-focused results with no suggestions for Apple, iTunes, or an iPhone-focused blog or website like Macworld, 9to5 Mac, or Macrumors.

Facebook’s basic results also let you drill down to a number of search refinements such as people, business pages, places, groups, Facebook apps, and events. If that doesn’t satisfy you, Facebook also offers links to Bing searches that, in the case of Moto X, were highly relevant to the query.

Google+ Search

Similar to Facebook, Google+ takes a granular approach to its search results. Searching for Moto X, for example, returns a jumble of all possible results focusing mostly on popular posts. You also have menu options to refine your search to just People and pages, Communities, Google+ Posts, Photos, Hangouts, Events, and posts and people from your circles only.

Google+ results take an “everything-but-the-kitchen-sink” approach.

Google+ offers a veritable smorgasbord of search results, but the search giant doesn’t seem to be applying the same sort of intelligence on Google+ that you see on Google.com.

Just as with Facebook and Twitter, Google failed the Moto X test and wasn’t able to suggest Motorola’s Google+ page as a relevant result when searching for the new smartphone. However, a search for Vancouver Canucks did surface that team’s page as well as the National Hockey League’s Google+ account.

A similar search for New York Yankees also surfaced the team’s page as well as Major League Baseball’s Google+ presence. Perhaps Google+ search is a little bit smarter than its competitors. But the best you can say for Google+ results is that relevant account results are hit or miss, as are Twitter’s.

Search on social networks is getting a lot better than the old days when the best Facebook could do was surface a few people or page suggestions before pushing you off to Bing. But each company still has a long way to go if they want their search products to understand the intent of user searches and surface relevant data beyond just simple keyword queries.

Follow TechHive on Tumblr today.

Subscribe to the Best of TechHive Newsletter

Comments