|
|
|
|
|
|
|
|
|
Seek and you may not find When Jacques Chirac gave a New Year’s address in the gilded splendour of the 18th-century Elysée Palace, political pundits were surprised to find him talking about search engines. “We need . . . to launch Quaero, the first truly multimedia search engine, to take up the global challenge posed by Google and Yahoo,” said the French president. The Franco-German project – whose Latin name, meaning “I seek”, has been chosen so as not to offend either country’s linguistic sensibilities – will focus on three areas: a combined sound, image and video search engine for the general public, professional search applications and audio-visual “heritage” such as historical footage. It is the first goal that has attracted the most attention – much of it derisory. “There is already a good French search engine – it’s called Google.fr,” jokes one blogger. The French satirical newspaper Le Canard enchaîné has also poured scorn on Quaero: “Compared with Microsoft’s €30bn [£20.4bn] profits or Google’s €100bn capitalisation, Chirac’s announcement is really going to spread panic in Silicon Valley.” True, any panic has thus far been taking place in Europe. The continent’s fears about its lack of progress in research and development, faced with the challenge from the US and Japan, are felt acutely in France and have played a part in the setting up of Quaero. French governments have a deserved reputation for adopting a dirigiste approach to technology projects and the broader economy, with mixed results. Two examples stand out. Minitel was a technological triumph when it was introduced by then state-owned France Telecom in 1983, allowing users to look for train times and pornography just as they do today on the internet, which superseded it in the following decade. But the system’s success was based on the offer of free terminals for its users, and it never took off outside France. Taxpayers propped up Bull, the information technology group, with billions of euros, but it failed in the 1980s to live up to the dream of a French IBM. Nonetheless, the government has further plans to counter what it sees as the growing influence of the Anglo-Saxon world view. This year it aims to launch CFII, an international satellite TV news channel, with the ambitious mission of competing with CNN. Yet there is widespread scepticism that its budget of €70m a year will be sufficient to make the channel a success. Quaero will be a test of a new approach to research and development driven by the Agency for Industrial Innovation, the brainchild of Jean-Louis Beffa, chairman of glass and ceramics group Saint-Gobain. AII will provide half of the funding on the French side, with companies led by Thomson, the French media services and equipment group, agreeing to match it. Discussions on the German side are ongoing, but it is likely that a similar arrangement will be put in place between the government and companies led by Bertelsmann. The scheme is not part of a push for “national champions”, however, as consortium members also include small and medium-sized software companies, national TV archives and universities. AII is founded on the principle that a partnership between universities and companies can drive innovation, but only if companies are prevented from restricting the flow of information for their own purposes. The French team is adamant that this approach to innovation is closer to US practices than one might think. “The US finances massively the R&D of firms through different agencies, sometimes linked to the military sector, with important civil outcomes,” wrote Mr Beffa in his report to Mr Chirac on French industrial policy that laid the groundwork for AII and Quaero, its flagship project. After all, Quaero members argue, the internet itself was born out of US military research. Members of the consortium that has formed to work on Quaero shrug off the criticism and also distance themselves from the political flag-waving. They believe they possess the right skills – and, importantly, a sufficiently large injection of public money – to make commercially viable technological advances. If successful, Quaero would indeed be the first “truly multimedia” search engine. Google may have created a powerful business around a good search engine but for indexing and searching for audio, photos and video nobody has yet won the battle, the researchers argue. “At the moment, I don’t think Google has surpassed us in terms of multimedia search technology,” says Jean-Luc Gauvain, a researcher at the French National Centre for Scientific Research (CNRS) and the project leader for the French university team. “It’s clear, though, that their resources are colossal.” The problem with indexing images, sound and video is that words are required to categorise and describe them. But this clouds the results, says Edouard Geoffrois at the research division of DGA, the French government’s arms procurement agency, another member of the consortium. “If you type ‘George Bush’ in something like Google Image Search you get photos of George Bush but you might get useless information too. If someone calls his cat ‘George Bush’ and puts its picture on the web, you can get a picture of his cat as well.” The quest for Quaero is for searches to “understand” audio, images and video without recourse to a written description. Image-recognition programs already exist: LTU Technologies, a medium-sized French software company and member of Quaero, supplies forensic tools to the US Federal Bureau of Investigation to analyse images from seized hard drives and match them against known paedophile photographs. There are audio programs, too, which can transcribe spoken words. At CNRS, the challenge is to refine these tools to improve accuracy and then to allow searches to plough through the transcription rather than simply look at file names. The technology will become more important as podcasts grow in popularity. The fad for downloadable audio files took off last year but their proliferation means it is becoming harder to sort the wheat from the chaff. Mr Gauvain says: “We have the ability to store everything, the problem is we can’t organise and index it. To do it manually would take millions of hours – ultimately it’s just not possible.” If all audio content on the web could be automatically transcribed then this process would be straightforward. Although indexing multimedia content is at the root of the programme, Quaero is not just an internet search engine. Thomson is interested in offering an in-built search facility on the set-top boxes it makes and supplying applications to its television and film company clients. Bertelsmann wants to create similar multimedia search facilities on various platforms. The third strand is to index the vast archives of footage that the French National Audiovisual Institute and the German Studio Hamburg have at their disposal. Tools currently available on the internet are not up to the task. AOL bought Singing Fish, an audio and video search engine, in 2004 but its results are not impressive. Out of only 358 results for a search for “Ariel Sharon” last week, the top find pictured a smiling Mr Sharon next to the disconcerting headline “Israeli Prime Minister Ariel Sharon is out of a Jerusalem hospital and eager to get back to work”. The footage of the now critically ill Israeli prime minister dated from December. Whether or not Silicon Valley is shaken by Quaero, it will be watching with interest.
|
>