What Shape Will it Take Tomorrow?

Over the last decade Internet has proven to be a vast sea of information and

we keep on looking for tiny drops of information that answers our queries.

Search engines have now become an integral part of our lives. Out of those,

Google search engine has become so common and popular that it has become a

synonym for search. Be it a student, researcher, professional or any other

person looking for information on the web, and he is bound to look at some

search engine or the other to begin their quest, and in most cases the search

engine will have to be Google.

We too had an unsolved question lurking in our minds, so we googled up to

find its answer. My query was, “What was the name of the school that Mahatma

Gandhi attended?”, and the 10 links that came up on the results page gave vague

information leading me nowhere. A few links referred to a school or university

named after Mahatma Gandhi, while some other led to Wiki pages where I had to

search the content to seek an answer to my query. The same query when run on

other search engines too didn't lead to any straight answer. The result is just

a guess that might be relevant or not as an answer to the query, as it is based

on keywords that are part of the query, which could have a direct answer. But

for response, I had to browse through various links. This has been the

limitation of search engines till this point; they provide guesses rather than

the exact answer to a user's query. And this is an area where things are

changing rapidly, with top search-engines vying to make users' search easier,

faster and more accurate.

If you compare the present search results to what you were five years back,

you will notice a lot has changed in terms of presentations and also relevance.

With the advent of Web 2.0, search is undergoing tremendous changes with the

influence of local, personalized and universal search. Even mobile search has

emerged as a new trend. Let's look at some of the areas where innovations are

happening in Internet search and what we can expect the search to be like in

coming years. This will show us what an ideal search engine could look like.

Knowing user intentions



The starting point for innovating in the search engine begins with understanding
the 'intent' with which a user approaches a search portal. There are a

considerable number of users who use a search engine for 'domain navigation; in

other words, to get to a site they have in mind. But a bigger chunk of users are

in one way or the other 'looking for answers' to a particular query. And there

is a limit to which technology can go in understanding the 'intent' of the user.

He is obviously not expected to be taught how to search, and thus the search

engine should be 'intelligent' enough to understand what the user might be

looking for. Top search engines are now trying to have the ability to know a

searcher's intent by knowing the mode he is in. As a searcher can be looking to

buy, to research or just be entertained, each of these different modes can lead

to a totally different result for the same search. Microsoft with their Live

Search and Google are focusing on knowing what mode users are in when they are

using search, so that they can be provided with the best results possible.

Apart from the mode that users are in, the search service providers are

trying to incorporate algorithms and AI techniques that help them determine the

intent of the query more accurately, despite common problems such as spelling

errors, punctuations and common words ( like 'is', 'are' that are mostly

neglected from the search phrase). Thus not just emphasizing on the keywords,

but also trying to evaluate the meaning of the search query. Hakia (www.hakia.com)

is one such search engine that uses QDEX (Query Detection and Extraction)

indexing model, which does a deep semantic analysis of the pages it crawls. It

not only matches the keywords from the search query, but also formulates the

meaning of the query to match with the pages' content to produce a relevant

result. Knowing the intent of the searcher is what can lead to provide more

personalized search results to the user; Yahoo-Mindset search is based on being

intent-driven.

Personalized search



Evaluating the intent from the searcher's query is one aspect, but what if the
search engine knows about the manner in which a user search for information can

lead to more accurate and relevant results. This is where all top search engines

are focusing upon. The search engines will keep track of your search trends,

your browsing preferences and also about your location. Of course, all this

personal information will be taken with the permission of the user. Google

Personalized search determines the location of the user from the IP address

where the query was generated and thus returns results that are from the user's

region.

The results on Google are personalized. It also provides options to promote or remove a result from the list and to add comments to results.

Based on the search queries that a user had used earlier, search engines will

be able to determine the context in which search keywords are to be used.

Suppose, if the query fired is 'LBW', then from previous search behavior of the

user, search engines will have the ability to server that intent by producing

personalized results of 'Leg Before Wicket' for a cricket enthusiast, while for

a medical student it will generate results that provide information on 'Low

Birth Weight'.

Localization of search



Since location is relevant in a lot of searches, incorporating user location
proves pivotal in increasing the relevance of the search. But the correct

meaning of localization won't just be limited to the location of the user, but

also in incorporating the language of the user. The web is a vast information

resource, and for every query there exists an answer somewhere, but what would

be the use if search engines could not produce the result in a language that the

user reads. Search engines are now eclipsing this language barrier by investing

on machine translation. A concept whereby the results will be produced to the

user, translated to a language he speaks and reads. Search service providers

like Google and Yahoo are providing an option that lets them to translate a page

into their local language. But the auto-translation of the result is what is

under research and in coming years we can see true localized search engine

results.

A visual search engine determines the similarity of image content through analysis of color, shape and texture.

Collaborating Media



You fire a search query and you get 10 links as result. Browse those links to
get the response to your query. Now imagine for the same query you get a result

page that not only fetches you 10 links to web pages but also shows some images

related to your search, news feeds along with blog reports pertaining to your

search query. This is what search engines are already transforming into;

collaborating information from various media into one presentation. Google's

Universal Search is one such example, which includes images, videos, books, news

and map information into the results page. This is a radical change from just

getting 10 guesses as result; now a user is presented a rich media which answers

the user's query. For example, a query fired about a song, will result in a page

that presents the user a video of the song, lyrics of the songs, along with

images and information about the artist whose song it is.

Vertical Search



Vertical search refers to a specialized search. Unlike generic search, vertical
search engines look up a very limited subset of the wealth of information in the

Internet. Such search engines are meant for specialized needs of users. Vertical

search engines send their crawlers out to a highly focused database to bring

back specific information as desired by a user. Such results are most value to

people interested in a specific area. However the companies that advertise on

such search engines reach a very refined audience. Hundreds of such vertical

search engines have sprung up these days. Retrevo.com is a site that helps you

browse and find electronics items. Now there are search engines for virtually

everything including job seekers, travel, patients and veterinarians.

Yahoo recently announced an expansion to the BOSS technology called “vertical

lens”. This allows developers build their own vertical search engines using

Yahoo's search technology, through BOSS. It gives developers the ability to

re-rank and control the presentation of search results, besides giving them the

freedom to incorporate the search results database within their own custom-built

search environments.

Yahoo Glue fetches confederated results from all possible aspects of search and presents them in a visually enhanced manner.

Beyond text-based search



We have been bound to type words for a search. Though new paradigms are coming
up in the search domain, which will not bound a user to text based search, the

system will be able to comprehend the query based on voice or image. Not only

voice, a user could be able to transmit an audio stream or pass an image to get

results about them. Transmitting an audio stream will be like a feature called

TrackID, that's available with Sony Ericsson mobile phones, whereby a song's

snippet is transmitted to a server to fetch back information about the song.

Working around the same concept, if a user says a few words from Lenin's speech,

he would be able to listen to the whole speech as result.

The image based search is already present with most of the search engines,

but is based on tags associated with the image and thus leading to inappropriate

results at times. Development work is underway to incorporate facial recognition

as a method for image search. For example when you try to find a watch that is

similar to the one you have a digital photo of, the result can lead to

information about the watch details and where a user can buy it from.

Speech Recognition and mobile search



Speech recognition technology is a milestone towards making man-machine
interaction more natural.

It offers maximum benefit for Internet search engines when it comes to users

of mobile devices. And the technology is soon going to make its entry in to

commercial products. Both Microsoft and Yahoo already offer voice services for

mobile phones. The 'Tell me' service of Microsoft offers information categories

like maps, direction and movies and Yahoo's 'One Search' also works on the same

voice recognition technology. iPhone search capability is not the only voice

offering from Google; the directory information service from Google announced

few months back has become a full fledged product. The service allows users to

ask for business phone and address information. As experts foresee, speech

recognition could be the next big thing in mobile search as it saves time. Also

the mobile phones will make the search more accessible. Not only mobile phones,

but even hand-held devices or devices in cars can be used to access search from

voice commands to get information about a location or similar stuff.

Localizing search like never before

We talked to Prasad Ram, R&D Head, Google India to explore Google India's research on plans of integrating voice responses, 'desi' search suggestions and more as part of their upcoming search offerings for Indian users. Google's philosophy of 'browser is the client' and not just an application to search the Internet has resulted in a lot of ideas specifically targeted which will mold into specific product offerings to be made available to Net users here. These include tools like Map Maker, which allow users to 'mark' places on extremely localized maps, which can be shared with the entire community of users. This is aimed at tackling an issue exclusive to the Indian context that locations are highly populated and each area has certain specialties that can never be determined and put down for use on the Internet. Add to this the fact that Indian users trust their community more in comparison to an electronic resource; Map Maker hopes to be a popular resource in the weeks to come. Keeping Indian audiences in mind, Google has formulated certain products that hope to overcome the most obvious problems that are exclusive to the area. These include the issue of 'Indian English' where words are spelt differently from universal nomenclature to yield meanings differently. A subset of this problem is Indian proper names that tend to have multiple spelling options in English. Various spelling suggestions which can be arrived upon by an 'intelligent' search technology, is currently being developed by Google in India. In addition to this, a 'stemming' application is also being created which tries to intelligently derive what you 'might be' looking out based on current usage trends and popular keyword combinations are also being integrated with this. To round things off, Google is actively planning to integrate voice into its core Indian search applications. Currently in pilot stages in Delhi and Hyderabad, the tool will allow the user to 'speak' the query, which will in turn be interpreted by the search database and results will be determined after understanding the accent and pronunciation of the search query. This will of course require to be tailored to specific areas, even states and cities. Hand in hand with SMS search and integration with local languages, Google hopes voice search of this nature could find takers, especially amongst first time users. Vishnu Anand

Enterprise search



The practice of identifying specific content across an enterprise by employees
has been there for long. However, it is only recently that it caught attention

with offerings from Google and Microsoft. Prompt decision making, responding to

a customer, or taking a new intiative; all require back up with comprehensive

data. This calls for a necessity to have easy-to-use and robust search engines

on corporate servers. Enterprise search takes search one step ahead as it helps

in spotting specific documents across enterprise in no time. Besides, in

enterprise search, unlike web search is language-based .

Search Appliance



The Search Appliance from Google can search up to 10 million documents using a
single box. It also provides a personalized search experience for end users.

Optimizing search technology for business uses, the search appliance offers

universal search across a slew of sources, such as, intranets, databases,

applications, and content management services. Among the many new features of

Google Appliance, you have Personalized Search, which allows administrators to

adjust search results for different user groups based on department; alerts,

where employees can subscribe to email alerts for topics of their interest; and

Advanced Reporting where you can view and export daily and hourly result sets,

top queries, etc.

Recently Google announced Search Appliance virtual edition for developers.

The free test bed enables you to test your apps and see how it all integrates

into your development environment without paying a single penny. You can

customize the search result page or even extract it for use on any other

website.

Enterprise search in Windows 7



Windows 7 is the successor of Vista to be released in late 2009. Windows 7 is
coming up with many new features for enterprise search. It clubs together the

search features of Windows Vista, SharePoint Server 2007, and many others to

provide users with improved search experience across networked and local data.

Libraries is just another way of accessing documents across different computers,

hard drives or folders in a single view. Libraries can also be created project

wise under which you can organize and search files lying across several

locations. You can search remote document repositories, SharePoint sites etc

from a single interface with Search Federation. Besides, the Search Scope in

Windows 7 enables administrators populate links on the start menu or even in

Windows Explorer. Such links make it easier to access data sources on the

network.

Conclusion



The innovations in the way we search for information is an on-going process, and
in coming years we will see more options. The answers to user queries would be

based on their preferences, and from the best available information media. That

would truly constitute an ideal search engine.