08 Sep 2008, Posted by Eric Karstens in Internet, 0 Comments

Of webwashers and enablers: Multimedia Semantics Conference in Crete

How to make sense of huge amounts of data? That was the leitmotif of this year’s Summer School on Multimedia Semantics, held from 1-5 September. Scientists, students, and media professionals from all over Europe gathered at the Mediterranean Agronomic Institute of Chania, Crete, for a week of seminars, lectures, and workshops. On 4 September, scientists and practitioners met for a common day of presentations exploring “networked media production and consumption in a convergent world”.

The day was opened by Barak Pridor, the mastermind behind the Calais project, who elaborated on the strategic motivation for this high-end automated annotation (a.k.a. tagging) tool that was acquired by news behemoth Thomson Reuters in mid-2007. Although Calais primarily serves the purpose of even better exploiting Thomson Reuters’ huge own databases in the domains of news, finance, medicine, and law, the company also made a conscious decision to open it to third-party users for free – and is actually even encouraging such users to try and make money off it.

The EJC’s web project manager, Arne Grauls, remarked in Crete that while many other efforts are still dabbling around with the issue, only Calais seems to have the potential to become the Google of multimedia semantics – unless, of course, it is suddenly overtaken by yet another surprise Google project. Content that has been “calaised”, as Pridor put it, will be much easier to find even if the search is performed in a rather fuzzy fashion, and denote its connection to other content items across domains and platforms.

Say, for example, that you are searching for information about a specific health condition. Aside from getting the appropriate medical results, you might also discover the legal implications of the illness, its representation in art and literature, history, recent news, blogs, and so forth, all on a global level. Thus Calais reduces the “unwashed web”, as Pridor called it, into a coherent and reliable data collection.

However, having ample amounts of information at one’s fingertips is nice, yet not sufficient. IBM’s business services specialist Frank Poerschmann pointed out that technological tools for the integrated analysis of data from different domains, business sectors, or entities within large organisations such as corporations or governments are basically there already. Yet the new challenge is to make sense out of the data and to align information analysis with existing management structures. This is where multimedia semantics might come in handy. Automated detection of meaningful interconnections between data can make human work easier and inspire new ways of looking at and interpreting information.

Poerschmann illustrated this with a prototypical cross-media company that operates free and pay TV channels, websites, and direct customer services (e.g., merchandising, mobile phone subscriptions, video-on-demand, etc.). Within the universe of a company like this, there exist at least four different models of monitoring business. Free-TV is measured by statistical audience size, a website by the number of actual visitors and page impressions, a subscription scheme by average revenue per customer, and all of the above can be analysed by their customer lifecycle value, i.e., the costs and revenues for acquiring and retaining a consumer during the entire time of relations with him or her. Semantic analysis can help consolidate these filtered views for the good of the company.

Still, there remains the question of who will benefit from the comprehensive semantic information processing to come. Will it merely help advertisers to better target consumers, improve business and security intelligence, or will it also empower the end users?

This might be the point where journalists come in. In the day’s concluding panel discussion, EJC director Wilfried Rütten coined the term “media competence enabler” in order to describe the changed role of the journalistic trade in times of the web. He said that future media professionals would need to apply their professional skills not only to tell a story or to provide the news, but to help others to find and tell their own stories as well. This would put journalists in the position of a kind of referee, negotiating the multitude of sources, providing trustworthy platforms and points of reference.

All that is a far cry from the old-fashioned journalistic self-conception, in which journalists are in a position of proximity to the powers that be. Where they pass authoritative information retieved there on to the general public. But it is still no reason to fear that journalists will be rendered redundant. Robert Freeman, in charge of video at The Guardian newspaper, emphasised: “Professional skills always, always, trump mere access to the tools.”

Other presentations included one by Meinolf Ellers, CEO of German press agency DPA’s multimedia subsidiary. He decribed how DPA seeks to alleviate slowing business with its core customers, i.e., printed newspapers, through providing full-service customisable solutions for regional newspaper websites.

Julian Kücklich, researcher at the London-based Press Association, showed how computer games can be used to spread news and knowledge and even provide the opportunity to simulate and predict future events. This surprising conjunction of two worlds which, according to conventional wisdom, could hardly be more distant from each other, opens up the much sought-for opportunity to reach today’s juvenile audiences with serious messages.

Much anticipated was the appearance of Clancy Childs, a sales engineer representing the often rather elusive Google, Inc. He explained some recent developments and current strategies of the data search giant, e.g., the custom search engine, facial recognition technology applied within Picasa photo albums, and the project “universal search”. While today Google delivers ranked search results separately for the categories web, images, videos, news, etc., the company aspires to integrate all relevant search results within one comprehensive ranking in the near future.