Semantic technologies are the future of web personlization
9 min read
15 September 2014
The future of the web is semantic - at least according to French tech startup Sépage, which specialises in semantic technologies for travel websites.
However the little known, little understood technology is still crossing the distance between science and business.
Real Business sought comment from Sépage on what this is, and how they’ve built it.
So, what are semantic technologies?
Semantic Web is basically a new way of structuring data so that machines can understand them.
For instance, if you’re seeking information about Paris, your computer doesn’t know what Paris really is. It doesn’t know if ‘Paris’ is the capital of France, a city in Texas or the Hilton heiress. With semantic web, not only would it understand the difference between these concepts, but it would also make sense of them: Paris is a city, which is the capital of France, which is a country in Europe, etcetera.
It’s just like when you go on a Wikipedia binge, going from one page to another by clicking on related concepts: an actor’s page takes you to a movie page, which takes you to an other actor’s page, then to a book, a writer, etc. With refined semantic technologies, this process is automatised and similarities between concepts in this knowledge graph can be found in milliseconds, like neurones in a human brain.
Now the size of this brain – the amount of structured data available – is still limited, but it’s growing fast, thanks to initiatives like DBpedia, a structured version of Wikipedia. Governments understood quite early the potential economic interest of these technologies and are funding several research projects.
In the UK, sir Tim Berners Lee’s Open Data Institute and the Data.gov.uk projects are taken very seriously at a state level. The same initiatives are happening in France and other European countries. All of this leading to more and more structured data available, and therefore more potential businesses.
When did businesses start using them?
These technologies have been around for less than 10 years, which is a short time for a technology to switch from public laboratories to businesses. The semantic web community is rather small, and is still very much limited to academic research, but a few researchers have crossed the line. Seevl (seevl.fm) for instance, a semantic recommendation tool for music was founded a couple of years ago by Alexandre Passant. Milan Stankovic, founder of Sépage is also a well respected researcher in the semantic web community.
So you’re a semantic technologies firm for travel, how does that work?
We basically combine different sources of public data to understand users’ queries (search performed of pages visited) and then find relevant travel alternatives in the knowledge graph. These alternatives are generally displayed in a sidebar, which record click through rates that are a lot higher than what can be observed with other marketing tools (Google Ad Display, RichRelevance), thus proving the efficiency of our customer understanding. Among our clients, you can quote Nomade Aventure (nomade-aventure.com), the French leader for trek and adventure travels.
What can these technologies provide a business?
Knowledge is the key. Understand what your customers want, speak their language, and you will be able to serve them the right content, at the right time. About 10 per cent of websites visitors click on our suggestions – that’s twice as much as Google ads or Outbrain, for instance. Most important, the first use of our product by early adopters records double digits click through rates – five to 10 times more than the players mentioned above = and up to 30% more conversions for these websites.
The beauty of it is that understanding is made possible with open web data, like Wikipedia and such. Most of today’s personalisation solutions are based on a statistical approach. They are heavy machineries, which requires huge amount of users’s data, often private data, to mix and analyse. A big advantage we have, because we use public knowledge, we can provide personalised recommendations to anonymous users, individually. Theoretically, a website with one single visitor could use our technology.
And who uses them?
Facebook recently released the Graph Search, which enables users to find information in human language and Google have the Knowledge Graph. Also Siri, the iPhone assistant, is partly based on semantic technologies. But while these types of use are more related to services, we are exploring ways to bring direct value by using these technologies in a marketing context. To our knowledge, we are the first to do that.
We believe the potential is immense. Most of today’s digital marketing approaches aren’t actually personalised, even though that’s what they claim ; comparing your basket to thousands of others and cluster you in groups of ‘similar individuals’ can’t really be called personalisation… With semantic web technologies, you don’t care about what other visitors previously did, and you don’t need to collect data. Every demand is analyzed individually and recommendation can occur from the first visitor.
In the future, this could also be used in other domains. Take the example of cultural items. Websites like Amazon already have recommendatory systems based on statistics (‘Users who bought this also bought…’). This type of approach works, but has the disadvantage of providing converging recommendations. For instance, if you buy a Star Wars DVD, the recommendations will probably be Star Wars Episode two and three. Now, with a semantic recommendation approach, the algorithm would search the graphs and find out, for instance, that Star Wars was inspired by Isaac Asimov’s Foundation books, or other similar science fiction movies. Much more inspiring. And inspiration can be very valuable [for businesses].
Where are you targeted toward, mostly France, or are abroad as well?
We have encountered strong interest in France but some companies in England have also expressed interest and we are planning on opening an office in London before the end of the year. We aren’t specifically targeting US companies today, but we surely will in the near future. We also have other travel-related projects coming up, like a search bar to perform natural language queries, for example if someone types ‘I’m looking for a cheap holiday destination near the beach, anytime this august,’ we want to answer that query.
We have big ambitions for the future, and semantic web technologies can go way beyond the e-travel market. Our goal is to create a true customer knowledge solution for companies, enabling them to understand their visitors’ behavior, the same way a human vendor would do.