If you are in a strange city and want to find a good local restaurant, the easiest way is to ask your hotel concierge. But for how much longer? The advances we have seen in mobile technology in recent years, particularly in areas such as natural language intelligence and location-based services, mean that mobile search just keeps getting better.
A new study conducted by Nielsen shows just how important mobile devices have become as local research tools.
The study focused on mobile activity across three categories: restaurants, travel and automotive. According to the study, which was conducted in the automobile-dependent US, cars are the most popular on-the-go location for mobile searches, with activity often spiking around highway exits as people look for nearby restaurants or other venues. Information like phone numbers, directions and maps were key for mobile users.
Among the three industry categories, users wanted the most immediate results when it came to restaurants. Almost two-thirds of smartphone users want to make a decision about where to eat within an hour.
Mobile search seems set for a bright future if it can deliver this type of local information effectively.
But still, I would argue that there is something missing, and that something is context, which is where natural language technology could really help improve mobile search.
Let’s go back to the restaurant example. You ask the concierge where a good nearby Chinese restaurant is. At the moment, mobile search engines understand “nearby” and “Chinese restaurant” and so can produce a fairly comprehensive listing of Chinese restaurants within, say 2km, of your current location.
However, mobile search engines do not understand “good” and why should they? “Good” is a value judgment made by humans and it has various implicit meanings. At its most basic, good is the opposite of bad. So perhaps you simply want to exclude restaurants where people have complained about the food. Or perhaps you understand “good” to mean that the restaurant has been given the most favorable write-ups by other users. Or perhaps you are used to frequenting Michelin-starred restaurants in which case the bar for “good” is set higher to mean restaurants that have been awarded some recognition or accolade.
Or perhaps “good” for you has a fairly precise monetary value: you want to spend about €40 a head, no more or no less, as that is the limit of your business expenses.
The hotel concierge does not know exactly what “good” means for you either. But he has a fairly good idea about what type of restaurant would suit you best from your age, the way you dress, the company you are with, the fact that you are staying in a four-star hotel, etc.
A mobile device has none of these clues and so is at a big disadvantage.
So could mobile search one day replace the human concierge’s extensive knowledge of the local restaurant scene and innate ability to infer your likes and dislikes?
Natural language technology is going to be the key to making location-based services a commercial success, I believe. Context-relevant information needs to go beyond the current focus on geolocation data if it is to discover the real “clues” as to what the user requires. Given the popularity of social media, social networks are a good place to start looking for those clues. Location-based services have yet to truly exploit this rich new source of shared data, although they are clearly moving in that direction.
For example, a mobile search engine has a greater chance of finding a “good” restaurant if it can search the postings of your friends in social networking sites. That has the added benefit of also giving you a fresh topic of conversation next time you hook up — “Hi John. How did you like the Golden Dragon restaurant? I believe you’ve been there.”