Blog

News, Disinformation & Language Intelligence

How Language Intelligence Helps Combat Disinformation and Supports Reliable News Journalism

Language in its written, spoken and visual form is the key medium of news content. The tools and technologies that have been developed to create and manage this content can therefore be called collectively “language intelligence (LI)”, a term that may be applied in a more general context and which is not confined to the media alone. The application of LI provides a critical mechanism across the digital world, where speed, data
volume, trust, cost and quality of knowledge all need to be carefully handled and balanced to deliver reliable products to networked customers on a global scale.

This report (“Orientation Paper”) attempts to characterize the various roles that language intelligence plays in the News value chain by asking the following three questions and providing answers to them:

•    Are there reliable automatic decision procedures for identifying the truth/falsity of news and similar content in general or in specific cases? Problem solution technologies.
•    Which available LI technologies can help news producers most to accelerate processes while lowering risk and complexity? Technologies along the News value chain.
•    What kinds of language solutions are needed and likely to emerge? The way forward.

At a workshop on 29-30 November 2018 in Vienna, Austria, co-organised by LT-Innovate, EUROSINT and SAIL LABS, various use cases that addressed the fake news problem were presented, and discussions into how Language Intelligence can extend its applicability within the news and intelligence industries in their effort to track down fake information were held.

This report is the result of the workshop, with contributions from all speakers and co-organisers. Editors-in-chief: Andrew Joscelyne, LT-Innovate; Margaretha Mazura, EMFS. Supported by the Vienna Business Agency.

Interested to join the debate? Come to the 8th LT Industry Summit  in Brussels, 24/25 June 2019, and discuss the topic in our session on: “Language Intelligence: Informing and re-tooling the media”. For more information, contact us!

DOWNLOAD THE REPORT

The Sound of Intelligence

This year’s Language Technology Summit, the 8th since 2012, is all about intelligence: human, artificial, language, and augmented (my preferred way of spelling AI).  In two days, the programme features the best of language technology (with the LTI awards at the end), vertical sectors (automotive, banking), looks into other interactive technologies like AR/VR – augmented and virtual reality, and showcases latest language intelligence developments for business needs.

The Summit is Europe’s major event showcasing the latest developments in multilingual ambient intelligence: speech interaction, deep meaning processing and multilingual communication & cognition. It is the ideal meeting point between technology providers, users & integrators, business leaders, leading-edge developers & researchers, investors & analysts. It is a unique place to learn, exchange, network, forge new partnerships, find new clients and get tuned to the multilingual intelligent future.

Join us on 24/25 June 2019 in Brussels !
Check out the programme & speakers: LT Industry Summit
Registration
Any Questions? Contact us!

Paradise Lost – the proposed EU Copyright Directive

Once upon a time, when we were dreaming of frontier-less information sources, we suddenly saw our dreams come true: The Internet emerged and with it, virtually border-less opportunities. It was paradise for creative souls but those who know their bible must have known: it will not last long.

The latest of proposed restrictions deals with the hotly contested Copyright Directive, or more precise, Art. 11 and 13 thereof. Art. 11 limits references to published news articles to single words or very short extractsotherwise a license will be required. First of all, it is not at all clear how short “very short extracts” are. Interpretation of these three words may vary considerably. But what is more, there may be a danger that “snippet” services as run by many associations as a service to their members (by their nature often not-for-profit) will have to be discontinued: It is impossible to get licenses (paying or not) from all news services that one may or may not refer to.

 Art. 13, though, is worse. Because it is definitely a restriction of free expression. It declares that commercial sites and apps where users can post material must make “best efforts” to pre-emptively buy licenses for anything that users may possibly upload – that is: all copyrighted content in the world.[Source: Blog of Julia Reda, MEP that is a MUST to read and to act upon!]

Example: You see the picture that I used for this blog entry. Who knows whether I am the copyright holder? Worse: How can I prove that I am the copyright holder? Even worse: Who cares for me, tiny little star in the Internet universe, to ask me for a license?

Consequence: There is a high potential that my uploads are blocked by a filter, just in case…much better to have problems with me (if I complain at all) than with the regulatory authority! [To reassure the reader: I am the copyright owner of both, object and photo, and the original artist is more than 70 years dead].

Therefore: Set a sign, join the 4+ million signatures and contact your MEP. Because it is not too late! All MEPs have to vote in the plenary (March or April, BEFORE elections). Maybe they get it right, eventually.

2019 – the Year of Interactive Technologies

The digital revolution is increasing its tools portfolio every year. But one concern is how to make the Internet and related technology areas more human-centric. That is: how to adapt and develop technology to become intuitive for human users. Human – machine interaction is on the forefront (think of chat bots, info bots or “Alexa” & Co.). Typing is still the main way to ask and communicate, but voice interaction in multiple languages will soon be taking over. Language Intelligence (Artificial Intelligence combined with Language Technologies) will solve accessibility problems (e.g. from minority language speakers), scale up data analytics and support the combat against ubiquitous disinformation (“fake news”).

Augmented Reality (AR), the enhanced real-world environments augmented with superimposed computer-generated images thus transforming and mixing real and virtual worlds, found its way into airports (e.g. Gatwick) and shops (e.g. IKEA). Its potential for entertainment is already well known, but work place AR is the future, soon to come.

2018 paved the way to raise awareness of these technologies and soon they will be  mainstream tools for the daily life. EMFS is proud to be at the core of these developments having facilitated the following activities:

Combat Disinformation: Check out the presentations of the “Fake news and other AI Challenges for News Media” – and watch this blog for the study paper that will be released soon!

XR4ALL – a Horizon2020 project that will develop immersive/XR (virtual, augmented and mixed reality) networks and platforms, and issue calls for sub-projects. Sign into their website and get timely information on how to participate!

The Sound of Intelligence: The 8th Language Technology Industry Summit, to be held from 24-26 June 2019 in Brussels, will feature latest developments of speech technology supported by AI, vertical challenges and cross-over sections, with participants from industry, research and clients/users. The Call for Speakers is open!

For more information, please contact us!

History and Future of Fake News – Vienna Event 29/30 November 2018

When Marie Antoinette went to the scaffold, she was not so much condemned for what she did to the French “citoyens” but for what she allegedly did, based on calumnious pamphlets: emptying the coffers of France for her whims and lustful, immoral entertainment thus committing high treason. Nobody believed her truthful protestations.
Rumours, misleading information are as old as mankind. For the purpose of profit, propaganda, advertising or simply upsetting the established order, fake news were propagated in an unashamed way throughout centuries. A comparatively harmless hoax in 1835, describing fantastic fauna and flora on the moon with some scientific claims, made the New York Sun the worldwide bestselling newspaper, overtaking the London Times. Anti-Jewish allegations had often disastrous effects, from medieval “Christian baby slaughtering” that caused pogroms all over Europe to the “Stürmer”’s caricatures and overt incitement to hatred.

That fake news appeared on the top 10 Issues to watch during 2018 (Disinformation & cybersecurity) does not come as a surprise, given the nowadays lightning speed of viral propagation of fake news on social media channels. This makes combating fake news more difficult. Researchers at Singapore’s Nanyang Technological University noted: “The difficulty in distinguishing fabricated fake news occurs when partisan organizations publish these stories, providing some semblance of objectivity and balanced reporting,” [Source]

Claire Wardle of First Draft News identifies seven types of fake news: [Source: Wikipedia]

  1. satire or parody (“no intention to cause harm but has potential to fool”)
  2. false connection (“when headlines, visuals or captions don’t support the content”)
  3. misleading content (“misleading use of information to frame an issue or an individual”)
  4. false context (“when genuine content is shared with false contextual information”)
  5. imposter content (“when genuine sources are impersonated” with false, made-up sources)
  6. manipulated content (“when genuine information or imagery is manipulated to deceive”, as with a “doctored” photo)
  7. fabricated content (“new content is 100% false, designed to deceive and do harm”)

But how to find out? European governments are quick in proposing rules and laws, but they will easier curtail the freedom of press and speech than efficiently combat the originators of fake news (the German example of its law against hate speech shows it, see e.g. article on Digiday ]

As Tom Standage of The Economist correctly stated: Media organisations and technology companies are struggling to determine how best to respond. Perhaps more overt fact-checking or improved media literacy will help.

One step in the right direction is the publication of the leaflet of the International Federation of Libraries Associations and Institutions IFLA:

How_to_Spot_Fake_News_CC BY 4.0 IFLA

To raise awareness about the challenges posed by fake news but also about the technical solutions to detect and combat them, LT-Innovate, Sail Labs and Eurosint organise a conference on “Fake News and other AI Challenges – News Media in the 21st Century” in Vienna, Austria, on 29/30 November 2018.

Join us in Vienna: Registration
More info or comments: contact

 

LTI2018 – Discover the Power of Multilingual Ambient Intelligence

The 7th edition of the Language Technology Industry Summit will be held on 28/29 May 2018 in Brussels. Target attendance is 150+, a mix of techies, CEOs, marketing people and business users, with special sessions involving the most recent insights into multilingual Artificial Intelligence, chatbots, hearables, emotional analytics etc. LTI18 is the place to be for LT providers and users, business leaders, technology developers, investors and analysts to forge new partnerships, find new clients and get tuned to the multilingual intelligent future.  “As a company we are regular participants and have benefited tremendously from participating both in terms of learning about latest developments in the language field, networking and meeting business partners” says Andrzej Zydron of XTM International.
Past speaker companies shows the exciting mix:

SpeakerCloud

Join LTI2018 in May – hot tip: Why not join LT-Innovate? The yearly membership fee is less than a full ticket and gives you many more benefits!
Not yet sure? Check out the past LTI2017 event or enjoy the happy faces on the photo gallery.

 

 

 

 

 

A splendid R&I future without bureaucratic burdens

They did it! I cannot believe it and nevertheless, it is true, black on white. A first attempt to really simplify participation in H2020 R&I projects. If continued in FP9, this would be a genuine game changer that guarantees (for the first time) inclusion and opportunities for all!
I mentioned it in the previous article but it might have been overlooked in the text: The last series of calls run lump sum pilots in 2 action lines (ok, not much, but “con algo hay que empezar”, as my first Spanish sentence said – we need to start with something).

Why a “lump sum project”? The Lump Sum Fact Sheet gives explanations: Funding based on reimbursement of incurred costs requires extensive financial reporting and is prone to error. Lump sum funding on the other hand provides considerable simplification potential. It will remove all obligations on cost reporting, and remove the need for timesheets and for financial ex-post audits on costs incurredremoving a big part of the administrative burden on beneficiaries. 

How would it work? The two pilots test two different options.
Option 1: The lump sum per project is fixed by the EC: Proposals describe the efforts and resources that the applicants commit to mobilise for the pre-defined lump sum amount. Applicants must also provide the proposed split of the lump sum per work package and per beneficiary. Proposal evaluation – and competition between proposals – ensures that an appropriate level of resources is committed. It will be tested in “Digital ‘plug and produce’ on-line equipment platforms for manufacturing” (IA) within the NMBP part of the work programme.
Option 2: Applicants propose a lump sum based on a detailed estimation of costs. This will be tested in “New anti-infective agents for prevention or treatment of neglected infectious diseases” (RIA) within the Health part of the work programme.

Why should EVERYBODY encourage the introduction of lumps sum projects in R&I EU programmes?
Did you read the most recent Annotated Model Grant Agreement? No? I do not blame you, it has 750 pages to explain the “simplified” procedures of ALL H2020 possible projects. Until you find out which do and which do not apply to your case, your project has probably ended. SMEs or start-ups may have a tough wake-up call when an Audit knocks at their doors and they see their personnel costs rejected as “ineligible” (I know about such cases). Ah, you say, I can always apply the lump sum rate for SME owners. Yes indeed, you can, but do you want to? The sum is fixed by the EC at a monthly EUR 4650 for 143 hrs for Belgium as 100% (or an hourly rate of EUR 32.5). This sum is then multiplied by a country coefficient, also given by the EC. This coefficient differs from 68.3% (Romania) to lucky UK with 120.3%. If you deduct from these amounts taxes and social security, you wonder why you should bother…

Check out the Lump Sum Fact Sheet for the beauty of the scheme.
If you are interested in a Joint Declaration to congratulate the European Commission for this idea, please contact me, I am happy to provide a text.

 

Horizon 2020 at its Zenith – new opportunities until the programme end

End of October, the European Commission presented the Horizon 2020 Work Programme for the last three years, with 30 billion EUR to invest in projects. The new features include measures to support market-creating innovation, highly integrated activities called focus areas, emphasis on better dissemination of results and a focus on open access to data.
Market creating innovation is the name for a first phase European Innovation Council, grouping together measures such as the SME instrument and Fast Track to Innovation, for enterprises, with a dedicated budget of 2.7 billion EUR. The SME instrument is now fully bottom-up, i.e. no thematic constraints. For phase 2 proposal, a hearing is foreseen, if the proposals pass a first stage.

The environment, the Digital Single Market, as well as the “Security Union” are among the priorities, so-called focus areas.

There is one interesting feature, labelled “second simplification”: Lump-sum pilots.

This new feature refers to a lump sum for the entire project, either pre-defined by the EC (option 1), or calculated as a budget and accepted by the EC (option 2). The beauty of it is: All is dependent on performance, no paperwork. While these 2 options are currently only available for 2 action lines (1 for each), this may be the solution for the future – and the successor R&I programme: Partners can focus on real work and forget about timesheets, proof of payment and the Damocles sword of financial audits (with rules sometimes unilaterally interpreted!). The EC should be commended on this initiative!

Call for proposals are accessible from the Participants Portal

A list of all calls (also old ones) according to acronyms is available here

ICT Work Programme 2018-2020

All Work Programmes

The latest version of the annotated model grant agreement shows all explanations – but needs some patience: it has by now 750 pages. Maybe this was one of the reasons why Lump-Sum Pilots were created…

Any questions? Contact us!

 

How “hot” is Language Technology?

MultilingualismImage
I am always amazed about the speed of changing trends in digital technology and their impact on language technology (LT). Some years ago, Big Data were the hype. In 2016, Hyon S Chu in his blog claims the death of Big Data (or at least the death of Big Data as buzz word). Exit Big Data, enter AI. Artificial Intelligence is the latest fad that will breathe brain-like abilities into machines. According to Forbes, the 10 hot topics of AI include (amongst others): place 1: Natural Language Generation; place 2: Speech recognition; place 10: Text analytics and NLP. AI will be a game changer, with most effects on vertical sectors (that already do or do not use AI).

These developments are in stark contrast to the average perception of people. Firstly, it can be observed that the term “language technology” is not “hot”. It is not even lukewarm. Reason for it is that LT comprises many different technologies that may (or may not) ring a bell. By now, people know Google translate, therefore, LT is associated with free, simple but not too reliable machine translation. Others may realise that the GPS voice might have something to do with LT (vocal human-machine interactions) but without labelling it LT, rather speech technology or similar. Secondly, when it comes to the most powerful segment of LT, text or speech analytics, the average non-techie person (in companies but also decision-makers) are at sea: How can LT have anything to do with defence or cybersecurity? How can LT cause a quantum leap in efficiency, effectiveness and security in the banking sector? Answers to these questions are indeed “hot”.

If you are among those that wonder or among those that know but want to hear more about future trends, join me at the LT Industry Summit 2017 from 9 to 11 October in Brussels where “hot” LT solutions for defence, the financial sector, business, and media & publishing will be presented. Some tickets are still available, register now!