In.Tech had a chance to chat with Google executives and the conversation turned to search.
GOOGLE’S success has been built on the “aha!” moments — that instant when someone types something obscure into the search box on its webpage and a bunch of accurate answers appear, ranked in order of relevance.
Because Google has made search so simple it’s often assumed that it has solved the problem of searching the huge mountain range of information that is the Internet with pinpoint accuracy.
But it hasn’t and it’s not easy maintaining the accuracy of its search engine, according to Alan Eustace, the company’s engineering and research senior vice-president. “In fact, it’s getting more difficult,” he told In.Tech at a Google convention in Tokyo.
"It used to be enough that we could return search results to users in the blink of an eye; enough simply to give users those 10 blue links to the answers they sought.
“But these days, user expectations have ratcheted up dramatically,” he said. Search engine users now demand more comprehensiveness, more relevance and even faster results.
Google has a set of programs called the “Googlebots” that continuously traverse the billions of webpages that exist on the Web, and the new ones that show up daily, or maybe even hourly.
This process, which it calls “crawling” is algorithmic: Computer programs determine which sites to crawl to, how often, and how much information to fetch from each site.
A Googlebot makes a copy of each of the pages it crawls to, to compile an index of all the content that it finds. Think of the “bot” as a cartographer, mapping all the content on the Web.
Through the index, Google can map each query to its page location on the Internet and then return the results to the users. And because the search is distributed across a large number of machines, it can perform the functions quickly.
It sounds easy. But when you consider the billions of webpages in cyberspace and that the numbers are constantly growing, you realise how easy it is for something to go wrong, and how easily an incorrect search result could turn up.
When Google started operations in 1998, it only indexed about 25,000 webpages; today, it indexes billions of them.
The human factor
Google also believes that successful web search is about providing the best search experience possible to its users. It has for years gone beyond algorithms to finetune its technology, enabling its users to help shape their own web experiences.
PageRank, for example, is based on the hyperlinks that people create around the Web. If Google believes that a newspaper site is more reputable than someone’s homepage, that’s because it is distilling all the links created by individuals on the Internet into one opinion about the reputation of each page.
The company also prides itself on its ability to go beyond simply counting keyword matches — where it would instead try to get a deeper understanding of its users, their queries and the pages they are most likely looking for, based on the way the users worded their search query.
This is where Google’s proprietary semantic search technology comes to play. In layman’s terms, semantic search is the idea that web search can be enhanced by “understanding” a user’s query, and any potentially matching documents, at a level that is deeper than matching the exact words in the query.
For example, when someone types in “Italian food” or “Italian restaurant” into the search box, any search engine can easily find pages that contain those words, but it takes knowledge of semantics to match a page that has the word “pizza” — which is highly likely what that someone meant.
What if the person asks, “Where is an Italian restaurant?” Not all search engines can process the natural, “normal” language people use everyday but Google can and will extract the correct answer, and list it at the top of its results page, saving users the time of looking through the results to find the best answer.
Google believes its search engine is faster, more comprehensive and more relevant than ever before. “But to carry on delivering those aha! moments on which our business depends, we need to close the gap between often and always,” said Eustace.
It now has many more Googlebots trawling the Web, guaranteeing up-to-date content from fast-changing sources such as online newspapers and blogs. And that’s also why the company has hired hundreds of scientists and researchers to make its search engine faster, more intelligent and easier to use.
“Our goal is to give all of our users the information they want, from the widest variety of sources, wherever they are, whatever language they speak, and whatever device they are using,” Eustace said. While Google can already provide some personalisation for a user, based on his or her location, users who open a Google account have the option of further personalising their search results.
Through iGoogle, users can customise their homepages. It lets them choose themes and gadgets to change the look and feel of their Google homepage.
Also, through an optional service called Web History, users can take search engine personalisation to a higher level. They can view and manage their web activities easily, get search results most relevant to them, and follow interesting online trends.
Google has not limited its search to text. Any query today pulls up information from all kinds of sources, such as books, videos, images, news reports, maps and more.
Known as Universal Search, it piggybacks off the variety of Google properties, including YouTube and Google Books. The idea is to bring various types of content to users through a simple interface.
It also sees that some users do not just search via a computer. Smartphones account for a high number of user queries. World Bank figures show that more than two-thirds of the global population live within range of a cellular network.
As such, Google offers various services on mobile phones; Google Maps, Google News and Google Web Search are some.
But it’s not stopping there. Eustace said Google is pushing search to the next level and beyond. Several of these initiatives include harnessing the power of the cloud computing, looking for better ways to search, and overcoming language barriers.
Eustace sees cloud computing linking up the world’s datacentres in a high speed network, which will enable hundreds of computers to work in parallel to process answers for user queries. “Search is going to get better as the cloud becomes more powerful,” he said.
Another exciting frontier being explored is search without having to type in words. This involves voice-activated search and image search. Google Goggles is an example of the latter.
It is also investing heavily in machine translation and transliteration, as well as machine learning, in order to make search better especially for non-English-speaking users.
“We’ve been working on Google Translate for years and now we’re extending its capabilities and reach to things such as a translation system that speaks back to you and auto captions on YouTube,” he said.
The future of search is about opening up the Web; to make all of that content more relevant and useful to more people, on more devices and in more languages.
“Quite simply, it means anywhere, anytime, wherever you want it, and however you want it,” Eustace said.