Nieman Foundation at Harvard
A news site in 2013 can’t be flat. That’s why French business daily Les Echos is gearing up to release LesEchos360, an aggregation tool that will cull articles from around the web to provide readers with an “expert filter” of business news. The project is now in private beta and is slated to be released in September.
Frédéric Filloux, general manager of digital operations for LesEchos — and maybe better known in the Anglophone world as co-author of the media and tech column Monday Note — said that he first had the idea for the project several years ago, when he discovered the popular tech industry aggregator Techmeme. “The idea of LesEchos360 is to give an overview of what’s going on in the economy by setting up a single unique homepage that will analyze in real time the business news cycle,” said Filloux.
Like the Lab’s own heat-seeking bot Fuego (now open source!), LesEchos360 will collect information from the Twitter accounts of Les Echos journalists, looking at who they are following and what stories are being shared on their networks. From there, it’ll rank articles according to a scoring mechanism determined in part by the number of retweets and the originality of the source, as Filloux explained to journalism.co.uk. He’s aiming to find sites that are off-the-beaten path: “The idea is to spot a very good blog, specialized on a very precise topic…we would like our system to be able to insulate and lift this stuff.” What they don’t want, he explained, is “a source which would be concentrated in the 20 best media companies and ignoring the rest.”
To make this all happen, Les Echos has partnered with French semantic analysis firm Syllabs, which specializes in web analytics and natural language processing. The project with Les Echos is an extension of the firm’s existing ventures in automated text generation and data mining, and the hope is that Syllabs will do for the paper what it promises to do for its business clients: increase visits to LesEchos.fr and make the site more adept at “informing and seducing” readers.
Syllabs cofounder and CEO Claude de Loupy explained that articles will be sorted reverse-chronologically into “clusters” according to their subject. “The problem is of course how to cluster — how to aggregate the different articles that are talking about the same subject,” said de Loupy. “For this we use semantic analysis. We want to do the same in English and other languages.”
de Loupy stressed that LesEchos360 is still “very beta,” and that they are still developing how the algorithm will rank and sort news items. For now, LesEchos360 relies on natural language processing to identify keywords identifying people, companies, or locations of interest.
“The semantic technology behind it is quite effective, but now the problem is how to clean the information,” said de Loupy. “You have a lot of links in the RSS feeds, so we have to clean these feeds, and I think the aggregation will be much better.” The beta version largely culls stories from major news sources, and a collapsible LesEchos360 top bar follows readers as they click through links. The page also features a sidebar of original Les Echos content.
“We went for a pragmatic approach,” said Filloux. “I wanted to do something which could be both an aggregator and a filter. My take is that we are fighting for time, we are fighting for the attention of the people, for their eyeballs — we need to make the experience of using the Internet and our content the easiest possible, in the most productive way.”
LesEchos360, like Techmeme, will not be fully automated, and is built to accommodate a “simplistic back office” for manual editing. Techmeme began as an automated aggregation service in 2005 and hired a team of real-live human editors in 2008.
As a news aggregator specifically for economic news, LesEchos360 shares some characteristics with Reuters’ Counterparties, the Felix Salmon-powered personalized link blog that he described to us in 2011 as a “BuzzFeed for finance.” The idea is the same: provide the most important stories of the day, as told by the best sources. But though the content of Counterparties is culled from an automated algorithm, the headlines are not; Salmon and fellow editor Ryan McCarthy isolate the most interesting part of the story and make that the headline, adding tags to each piece. This makes for a curated page of witty, sometimes even funny, straightforward content, and Counterparties has been able to experiment because money was never the end goal. For LesEchos360, that’s not quite the case.
“This is going to be a traffic driver for us. Since we have by far the largest business newsroom in France, because we are the only business newspaper, we [are] automatically very well placed,” said Filloux. LesEchos.fr attracted 3.7 million unique visitors in April 2013, according to Nielsen.
“It’s not a philanthropic project. We are going to run some advertising on it,” said Filloux, explaining that he hopes to be able to sell the semantic extraction algorithm driving Les Echos 360 to businesses looking to increase the profile of their home pages. “This won’t bring a huge amount of money, of course. We want to make an [example] of this technology — basically we want to be able to tell the business community, and the companies we already work with, that for your specific sector we can build a machine that will extract the relevant info to your sector.”
One of the main reasons for developing LesEchos360 is that the aggregator is a precursor to larger changes to the paper’s digital presence. Filloux said that a redesign of LesEchos.fr slated to debut in 2014 will allow readers to create a personalized “dashboard” of news items according to their interests. Readers will be able to choose whether they want to see the news through the 360 lens or not. Les Echos also recently announced that it’s working on a semantic enrichment project with Temis which will make it easier to cross-link Les Echos content and extract article keywords. Filloux said the Temis partnership is aimed at “re-assigning some value to our stock of one million-plus stories by Les Echos people over time. We are going to put our stack of stories in the system, extracting names of companies, geolocation, names of people, all sorts of codes.”
If all goes as Filloux hopes, it could be a heartening experiment in how text generation technologies could help journalists get more out of their original content, and potentially increase the premium on original news analysis. Whether it will be monetizable in the way Les Echos imagines, though, depends in part on how much companies would like to see themselves as providers of industry news. It’s hard to quantify exactly how much value semantic enrichment and analysis can add to news organizations, but as the French say, il n’y a pas de fumée sans feu.
Equirectangular panorama — 360 degrees, get it? — from the Pont de Bir-Hakeim by Alexandre Duret-Lutz used under a Creative Commons license.
