Semantic search – can it really pave way for a more efficient online world?
Setting real expectations from semantic search may be harder than we think. Even if a search query can be handled the way travellers think, it doesn’t necessarily follow that accurate results will lead to a shorter buying cycle. Ritesh Gupta investigates
For anyone who starts planning a holiday with just a vague notion of what they want, working out a trip online can be daunting. This also explains the reason behind the mushrooming of start-ups in the travel planning area. To be precise, it’s about targeting travellers in the early part of the booking funnel.
As for users, they might know what they want to do (a trekking holiday that works for kids, for example) but be lost on where they want to go. It is in this type of scenario that semantic search aims to help.
“I believe that semantic search technology has the power to really change the way travel is searched and consumed online,” says Saurabh Srivastava, VP marketing and product strategy, ixigo. Progress is definitely being made to improve the accuracy of results by intelligently deciphering what people are actually looking for, and the context within which they are looking for it. However, it still is far from perfect and will continue to grow and develop with more consumer data and actual travel query behaviour filling current gaps.”
Tudor Coman, co-founder of flocations.com, a travel start-up in Singapore, says judging by the overall popularity of semantic search, he isn’t surprised the travel sector hasn’t made much progress. However, there are interesting offerings such as adioso.com which promises the best flights to anywhere in one search. The team behind Adioso (founded in Australia in 2008) says the website helps by having “the best vacations humanly possible”.
Coman is clear about what he expects when he is searching; he expects to ask questions in the English language and get results that relate to his query regardless of how convoluted my question is. “I have a family of four and we need tickets under $500 each to Florida. We also need to stay at a five star resort with a trip advisor rating of five… If you can take that as an input, that’s semantic search.”
Benefits and challenges
According to Srivastava, the main benefit that semantic search offers over other ways of handling user queries is that it understands a user’s intent rather than simply aggregating results by a narrow keyword. For semantic search, this is where the real challenge lies.
“Across various researches and use-case analysis, the general consensus is that instead of having to piece together different parts of the holiday in individual search boxes, it would be preferable to have just one search box into which you could type your request,” he says. This has been the backbone of iXiGo’s semantic search box and over time the company has seen user behaviour of searches changing as they get more comfortable with this new paradigm of search and start trying natural language queries such as ‘where can I find street food in New Delhi’ or ‘best steak in Pune’ or searches that are used to make destination decisions, such as ‘beaches in India’, ‘hill stations around Delhi’.
Gauging performance
So considering all the talk about understanding a web visitor’s true intent and the contextual meaning of content, how does technology manage to return a structured search query with high precision?
According to Google, buyers peruse about 20 sites on average before making a final purchasing decision. “Visiting many sites is human nature and not a result of using a bad search system,” says Srivastava. In fact regardless of how accurate semantic search is or how relevant the results are to a user, they will still take any result and try to replicate it across many websites to ensure they are getting the best deal. For this reason, he does not believe that accurate results lead to a shorter buying cycle. An important point for sure.
Looking moves by Apple and Google
Countless startups and think tanks that have worked on semantic search which could be down to humans’ obsession with getting computers to interact with us as one would with another human.
“Apple’s Siri does an alright job of interfacing with the user and that’s because it is a spoken interaction which makes sense to have semantics,” he says. However, if you are interacting on a keyboard and screen, semantics search loses its charm. That’s why Google is still keyword search - humans interact much better using keywords rather than semantics when the interface is a keyboard and screen. However, if it’s a spoken interface with Siri, doing keyword commands would be awkward and embarrassing to perform.
According to Coman, semantic search comes down to two things:
a) Interface - if it’s a keyboard and screen, users are way better at navigating through menus, filters and keywords, rather than typing out a well thought out question. If the interface is voice, then saying what’s on your mind and having an interaction as you would with a travel agent will be more natural.
b) Preference - let’s face it, not all of us are articulate and many people don’t know how to ask questions.
“Therefore, if semantic search is just as accurate as keyword search, it will strictly be a user preference thing,” says Coman.
So why then is it still considered important? For one, Indian domestic tourism continues to grow apace. Says Srivastava: “We are poised for a very interesting phase. With people opting for travel at short intervals in a year as opposed to one or two vacations, and technology playing an ever-important role in the travel industry, users are constantly looking for quick solutions to their queries.”
With robust growth it will become imperative to provide relevant information and options to travelers, and this is where semantic search will play a crucial role in helping to quench the thirst for travel queries online.
With travellers able to use the search box as they would a real life travel agent or friend – asking vague or specific questions in natural language, companies like ixigo believe that they will be able to provide a far more personalised, satisfactory service. In turn, travel businesses will benefit, as customers are able to hone in on the hotel, restaurant or activity that perfectly suits their requirements.