Privacy on search engines not an issue

Though other components of what was generally termed web 2.0, such as blogs, wikis and social networks, developed over the course of the decade, search – lead by Google’s stunning growth still occupies the crucial position amongst them.

Even from those early days, it was clear to many search providers that an improved search experience comes with considerable accumulation of data that allows search engines to get a good idea of what an individual user has been looking for.

“Search engines ought to monitor users in order to understand who they are, what their preferences are and what they look for on the internet,” said Yoelle Maarek, Senior research director and head of Yahoo research in Israel.

“Users constantly enrich the efficacy of a search engine through their constant interaction with it, thereby enabling the engine to understand their precise needs. Their contribution is important for search engines, who look both to understand them for a better service and also use them to see if new features are to their likings.”

The Haifa-based researcher spoke last week in Bangalore about the “surprising” contribution of users in making web retrieval relevant. By clicking on a result or even leaving it aside, users are voting for or against a research the search engine tries to conduct without explicitly making it known to them.

Given its nature, researches in search often revolves around two important data gathering mechanism –query logs and click through — which enables developers to constantly tweak their algorithms and provide better results for the queries.

Users are being monitored based on the pattern of their queries, the sequence and the gathering of data over a period of time, which allows the engine to get a better idea of the information being searched.

Users monitored

“There’re a set of things that helps us monitor the user – the geography, the time and the sequences of queries. If someone searches for hotels it’s usual to find multiple searches either for different hotels in an area or associated queries indicating they may be working on a travel plan. There’s the sequential similarity in this dialogue process. We analyse and see if the sequences are always the same or if it changes,” Maarek said.

This also helps researchers understand a particular term or name users may be searching for without necessarily typing in the right spelling. “Britney Spears has hundreds of different types of spellings… We understand the name with each wrong spelling,” Maarek joked.

Sometimes, if the user searches with his/her log in information and has a similar kind of searches, it becomes easier to predict what they look for. Search engines also rely on the constancy people have in their requirements. Users, for instance, don’t change things they like or need in music, entertainment or things related to work. Longer monitoring will also provide glimpse of other activities such as tours they take, offering  search engines a good idea of what often prompts them to put out a query.

“Pinning users to a specific geography makes it even easier in terms of providing them the direct source such as sports, movies or, like in the case of those in Europe, transit information,” Maarek said.

“Users are generally happy when such information is served instantly. The very fact that they click something or stop putting out fresh queries are indications that the purpose of their search is fulfilled.”

In reality, web-based services have various means of identifying and learning about users. While search may enable them track users through their dialogue with the engine, browsers would provide even a wider window of what users do on the internet.

Those like Google has the convenience of having both the search engine –with 86 per cent market share worldwide -- and also the Chrome browser. The bits and pieces of their activities on the net often easily jell together for researchers to analyse.

“There’re three big challenges for search engines,” Maarek said. “How do we personalise the service through the large amount of data? Analysing a large data pile may help derive a generalised pattern and user may not find their rare needs met so effectively.”

Breach of privacy

Monitoring and “learning who they are” may help search engines serve individual users better, but Maarek acknowledged that it is important to draw a line between personalisation and privacy. While users may be happy to be given the best results, they are uncomfortable when the engine knows too much about them. “So drawing the line is important,” she reiterated.

Despite collecting data about users and monitoring them, threat for their privacy seems significantly less in search. “Breech of  Privacy is a bad news for our business,” Maarek said.

“When my teenage daughter searches for things, I know she won’t come across that many explicit things and we know how to keep that away. But as a mother, I’m more panicked when she comes in contact with an unknown person on the social media who may be an adult masquerading as a teenager. Frankly, such a scenario is more dangerous than an inappropriate search query.”

Fighting to retain users is another big challenge for the engines. “They won’t switch (to other engines) if they are happy with the results. So we work to keep them happy for which our research is important,” Maarek said. “In this aspect, I feel the Yahoo-Bing partnership is important. We have the data and also the features on our search engine and Microsoft pays for our servers and data centres to be on. This is bound to make the search engine more powerful and competitive.”

Getting personalisation right as the internet switches platforms from desktops into mobiles and tablets is crucial for the search to sustain itself as the internet transitions itself from typical desktop/laptop environments to mobile phone and tablet PC devices.

“Thaïs’s kind of easy as mobiles are often personalised devices for people. We’ve to make search implicit –delivering answers without the users querying about things,” Maarek said.

More than a decade on, search has matured into a fairly reliable means of finding information on the internet. Little wonder then that search engines like Yahoo now believe that they can understand the user well through his past searches and provide him the best service.