Google Improving Mobile Search
Making it easier to make entries on a mobile device
As smart phones, and web-connected PDAs become more and more common, it makes sense for search engines to consider how to make it easier for people to use those devices while searching the web.
Handheld and mobile devices can be difficult to enter queries into.
Google has come up with some ideas to make searching easier on a phone or PDA.
They involve a type of auto complete and spell checker that can predict what you might enter before you finish typing, or tapping, or speaking.
This type of predictive data entry gets its information from dictionaries that can be stored locally, or remotely. Those dictionaries can be built using information from a collection of email, or corporate documents, or from web searches.
The data in these dictionaries might be based upon previous entries, or the number of times words occur in those places, or occur together in the event of phrases being used. They might also include information about words searched for based upon the popularity of searches or recent searches.
In a previous post, I asked Can Google Read Your Mind?
That post was about a patent application that Google came out with that seemed to describe Google Suggest. But it also talked about a number of ways a search engine could process queries so that they might seem faster, and that would have a significant impact on how hard the search engine might work.
Google has released another patent application that also talks about “predictive textual outcomes” based upon search popularity, and using data dictionaries.
It’s worth considering carefully because it may be a viable way for search engines to proceed in the future, especially in an era that will likely see growing use of handheld devices.
The inventor, Shumeet Baluja, has written some papers on search, including a poster on mobile search titled Mobile Search: A First Glimpse Into the State of MobileSearch (pdf), which has some intriguing stats cited on it about the differences between mobile and desktop searches (for instance, the typical query used on a mobile device is longer than on a desktop).
His short, but interesting The Happy Searcher: Challenges in Web Information Retrieval covers a lot of ground, and is also worth a look.
The patent application is:
A computer-implemented method of providing text entry assistance data comprises receiving at a computer information indicative of predictive textual outcomes, generating dictionary data from the received information; and providing the dictionary data to a text entry device remote from the computer.
The received information may relate to search requests made by a plurality of remote searchers, and the received information may be indicative of search term popularity.
United States Patent Application 20050289141
Inventor: Shumeet Baluja
Published: December 29, 2005
Filed June 25, 2004
This patent application is aimed at helping users of computers or handhelds enter information by providing and updating dictionaries to remove ambiguity in text entered by those users.
The issues that this application addresses?
There’s a large growth in mobile computing, with an increased use of handheld devices. And a demand that those function as well, or maybe even better than desktop computers. That portablility demands a small size, and a limited means of entering information into the devices. Some handhelds have evolved shortcuts to letting people enter information, but those methods are limited.
Even users of desktop computers may want help with data entry or such things as spelling and grammar checkers.
Is there something that can help people enter information more quickly and accurately in both types of systems?
If we look on the Google Mobile Search page, it used to mention spell check a couple of times. This patent application defines how that spell check, and other methods might make using a handheld easier.
Viewing Information on a Mobile Screen
While the Google patent application briefly discusses what someone might see on a display screen, that isn’t the focuse of their invention. But it is something to consider, and this was one of the more interesting documents I found on that subject.
Written in 2002, this Microsoft paper discusses:
- Users’ difficulties interacting with mobile devices because of how information is input into the devices.
- People adopt a slower, more laid-back approach to searching on handhelds because of power limitations and slower connection speeds, and look at less search results than they might on a computer.
- People focus more on answers to specific questions on a phone or PDA than on advertisements.
- Location information tends to be more important on a mobile devices.
The document focuses upon the presentation aspect of serving pages to viewers on smaller screens, and segmenting pages into blocks, while determining which of those blocks are important enough to show viewers.
Might there be some important lessons there that could apply to a search engine deciding what content is the most important on a web page, and deciding that information might be the most important when relevancy in search indexing is determined?
Improving mobile search could improve desktop search
There are parts of Google’s patent application that could also be applied to desktop computing, such as the creation of data dictionaries for organizations based upon documents on their intranet or emails, to assist people in completing queries.
Some of the constraints created by the limitations of mobile computing may make desktop computing a more pleasant experience. Solutions like the ones described above from Google and Microsoft will play a part in the evolution of search, and how we use the web.