October 15, 2014

Google Now: Comprehensive, Comprehending

By Megan G.

Introduced as a mobile device app in 2012, Google Now provides “the right information at just the right time,” meaning the app uses different sources of information from your mobile device activities to specifically tailor Google Now updates to your interests.  For example, a person’s Google Now ‘cards’—individual tabs that show up in the app—may include commute traffic, local hotspots, or your favorite sports teams’ scores.  This all depends on how you use your phone and what you let Google Now access.  It gathers data in the background of your phone as you use it, and predicts what info you need, and when you need it—hence the tagline.  Google Now is a pocket-sized secretary in the same vein as Apple’s Siri and Microsoft’s Cortana.  Why label this particular app both comprehensive and comprehending?   Two major traits set Google Now apart from its competitors in the digital arena:

1)  Google Now pulls information from the ever-evolving Knowledge Graph of Google, transforming the app into a mini-encyclopedia, in addition to its standard search engine services.  This makes it comprehensive.

2)  Google Now’s natural language processor allows the app to comprehend the user.  Through conversational search, Google Now can understand spoken queries, and can answer many of the user’s related questions in succession.

An Impressive Knowledge Base

A recent study conducted by Stone Temple Consulting showed that Google Now ranks the highest in a contest of digital assistant knowledge boxes.  A “knowledge box” is the graphically-defined box that sometimes pops up in SERPs (search engine result pages) to directly answer a user’s query.  For example, if you type into Google, “define: lockbox” a noticeably raised box appears at the top of the page providing various definitions for the word.

No need to click on Wikipedia or Merriam-Webster to find your answer.  Google Now’s knowledge box got the job done, and according to the Stone Temple Consulting survey, it can do that job better than anyone else.  In a tournament of 3086 challenges, Google Now managed to completely answer 88% of the queries posed.  Siri came in at 53% and Cortana at 40%.   Google Now’s knowledge base is the most comprehensive of the three, thanks to Google’s Knowledge Graph.  The Graph expands in both its knowledge and understanding as people use Google.   It learns how to better answer user queries, and links certain searches together to actively predict user’s future questions.  Clairvoyance seems a shared quality among Google services, with the site’s AutoComplete feature offering possibilities, and modifying those possibilities, for each word the user types.

Stone Temple Consulting clarified that their study was not a contest of digital assistants as a whole, but instead one of knowledge bases.  So, what does Google Now’s comprehensive knowledge base mean for user experience?  Siri and Cortana more often route users to the correct answer, instead of providing a direct answer through a knowledge box.  Routing people elsewhere results in more laborious sifting and searching that can be avoided with more efficient knowledge bases.  More sophisticated natural language processors also help to expedite searches, which brings us to our next point of investigation:  anyone with a Smartphone has felt the frustration of a digital assistant misunderstanding a spoken command.  Google Now’s conversational search paves the way for an easier, more fulfilling digital assistant experience. 

Conversational Search

As described in Danny Sullivan’s Search Engine Land article, Google Now remembers the previous queries of the user, and does not treat each question as a separate interaction.  Sullivan has a chat with Google Now about Barack Obama’s age and height measurements—a rather dull conversation, perhaps, but the topic is not the point.  The cool part is that Sullivan did not have to remind Google Now of what they were speaking about.  Consider this fragment of their interaction below.  DS denotes Danny Sullivan, and GN stands for Google Now:

            DS:  “How old is Barack Obama?”
GN:  “Barack Obama is 51 years old.”
DS:  “How tall is he?”
GN:  “Barack Obama is six feet one inch tall.”   

You can find the rest of the conversation in Sullivan’s full article.  This interaction proves interesting for two reasons.  First of all, Google did not route Sullivan to any other website—not Obama’s official website, nor his Wikipedia profile.  The application already had the information in its knowledge base.  Second, Google Now understood that “he” meant “Barack Obama.”  Sullivan did not have to battle with the short-term memory of the assistant, and constantly repeat the subject “Barack Obama” as if he were talking to Dory from Finding Nemo.  Google Now simply kept up with the conversation.  It actually comprehended Sullivan, and their extended interaction. 

Whether or not Google Now is the best digital assistant remains to be seen.  It has the same hitches as other synthesized-voice assistants.  However, the recent study from Stone Temple shows Google Now leaving the other assistants’ knowledge bases in the dust.  And its conversational search, which understands that “pizza place” means “pizza restaurant,” can understand natural human language in a way that favors long-tail keywords and whole phrases—in other words, organic sentences.  Interacting with Google Now is a more satisfying, natural experience rather than a detached, virtual one.  

No comments:

Post a Comment