25 November 2013

OECD chooses TEMIS to semantically structure its Knowledge and Information Management Processes

TEMIS, the leading provider of Semantic Content Enrichment solutions for the Enterprise, announced today that they have won a call for tender issued by the Organisation for Economic Co-operation and Development (OECD) with their award-winning Semantic Content Enrichment solution Luxid®.

"This mark of trust by OECD represents a new recognition of our ability to address challenges in international organisations. For TEMIS, this is a link between our know-how in the publishing domain and our industrial experience in information systems", said Fabien Gauthier, Sales Director, Enterprise, TEMIS.

The OECD provides its expertise, data and analysis to its 34 member governments and 100 other countries to help them support sustainable economic growth, boost employment and raise living standards. To fulfill its vision of increased relevance and global presence, the OECD has launched a Knowledge and Information Management (KIM) Program that establishes an integrated framework for managing and delivering information and improving its accessibility and presentation. The KIM framework is intended as the steward of the OECD's information lifecycle, with a universal knowledge referential at its core facilitating enhanced searching and findability, rationalized content re-use/repurposing processes, and supporting the organisation's Open Data and Linked Data initiatives.

Based on patented and award-winning Natural Language Processing technologies, Luxid® exploits off-the-shelf extractors called Skill Cartridges® to extract targeted information from unstructured content and semantically enrich it with domain-specific metadata. This enables professional publishers to efficiently package and deliver relevant information to their audience, and helps enterprises to intelligently archive, manage, analyze, discover and share increasing volumes of information.

The full press release.

15 November 2013

Deep Multilingual Semantics Pays Off for Bitext

The Spanish semantic technology company Bitext (The Bits and Text Company) has been in the news quite a bit since it won an LT-Innovate Prize in June 2012. Analysts have complimented Bitext on its decision to develop a technology that enriches and extends existing systems with deep linguistic knowledge, rather than reinventing the wheel for each application. 
We caught up with Antonio S. Valderrábanos. CEO & Founder and Enrique Torrejon, the R&D Director, to check on their progress.

International strategy

Since June 2012, there have been three major events in Bitext’s business activities: In the last quarter of 2012, they reached an agreement with Salesforce, the leader in social media monitoring, to provide multilingual sentiment analysis in the Salesforce Marketing Cloud Insights ecosystem. With Bitext's real-time, multilingual sentiment analysis companies can understand their customers better and faster than ever before. This decision to join an existing sales channel seems typical of the company’s approach to market outreach.

They also signed distribution and partnership agreements with companies such as Actuate. This means that Bitext is steadily positioning itself as a text analytics and sentiment analysis provider for Big Data in the US market as a whole.

Here in Europe, Bitext is now working with the Spain-based telecoms company Telefonica to provide multilingual text analytics for voice of the customer in different languages for their international product launches.

Verticalising the technology 

In the immediate future, Bitext’s business agenda for the next three years is to build a stronger presence in the US – especially in Silicon Valley – so that it can sign partnerships with major US corporations. On the technology front, they will be focusing on “verticalising” semantic applications beyond sentiment analysis. This will involve developing text analytics for specific purposes, such as making recommendations, intent to buy, optimising contact centre performance, and also fraud detection.

How about the European market? Bitext agrees that one of the company’s major assets is the extensive multilingual capabilities of their solutions & services. But there is also a certain disadvantage in this for many European language tech companies, say Bitext’s senior executives: “Paradoxically, the existence of multiple languages in Europe segments the market according to languages. This makes it difficult for language technology providers to expand to other markets if they lack these multilingual capabilities.

What Bitext would like to see more of is better access to financing at European level, be it through business angels or investors in general. Even their competitors would probably agree with them on that point!

10 November 2013

Out of the Mouth of Babies: Young Europeans Should Learn to Code for Natural Language

Coding is becoming cool again. Western countries have counted the cost of failing to educate the next generation of ICT-smart young people. There was a drop in academic interest in “computer science” a few years ago and as a result there’s now a steady stream of government proposals for boosting information and communication technology education for the youth of European and other countries.

In September, European Commissioner Neelie Kroes and the EC Education Commissioner released some alarming figures about the state of ICT in the educational infrastructure and in terms of educational content. They detailed a list of 24 new initiatives to beef up ICT in education in Europe.

Digital jobs
In the United States they reckon that 1.4 million jobs—and 60% of sci-tech/engineering jobs of the future—will require computing skills. So if young people will have to become IT-savvy for almost any job, one way to kick-start at least some IT education at school would be to work from what children use most but know about least – namely, natural language. Using NLP as an educational problem space might offer an intelligent entry point into the great instauration of computer coding. At the same time we could help expand the community of NLP-aware coders and perhaps learn new tricks from a new generation of device-happy young innovators.

Coding cohorts
There have been courageous attempts to boot up a cohort of young coders in Europe. The Irishman James Whelton set up CoderDoJo to teach kids to code outside of the standard education system. 
Another much-praised venture has been the UK’s Raspberry Pi 'simple computers for kids' project, driven by Cambridge University stakeholders in the UK. In late October this year, they notched up their millionth computer manufactured in Wales - a credit-card sized device that plugs into a TV and a keyboard. The founders wanted it to be used by children all over the world to learn programming. But it actually may be more popular among older IT hobbyists, as this news item about building a speech translator suggests.
Can these efforts really rise to the challenge of fostering the kind of large-scale IT literacy being proposed in countries from the US to India? In the US, the Association for Computing Machinery is holding an Hour of Code to introduce more than 10 million students of all ages to the basics of coding - “a foundational skill for careers in the 21st century”.

Hackathons, Code-Ins & Community Building
It is heartening to see that Europe’s Apertium free/open-source machine translation community is already participating in this year’s Google’s Code-In. The idea is for students from all around the world to tackle small tasks (code writing, debugging, documentation, production of training material) to learn how to prepare for larger projects in the future. But it’s only a start.

Europe’s multilingual footprint poses a tremendous challenge for cross-border transactions. But there are already hackathons (e.g. the Moses Marathons for statistical machine translation) that can help open up opportunities for dedicated communities. 
So let’s give more incentives to younger hackers with an interest in the world of apps and human language, and encourage them to learn about existing resources and APIs to create new ways of addressing our language needs. Multilingual communication ought to be an enjoyable, cheap and profitable challenge. 

05 November 2013

[Presentation] Language Technologies and Business in the Future

Very interesting and valuable presentation by Niko Papula, from Multilizer 
at KITES Symposium, Helsinki 31.10.2013

Exalead: very recherché

With its “connect the dots” tagline, Exalead is one of Europe’s most successful search companies to emerge in the last decade or so, and is ranked fourth among the world’s most-used web search engines. Largely focusing on the strategic business of enterprise search, the company has nevertheless explored the whole search space, from desktop and web search through to multimedia and voice search in some of its Lab projects. We see three good reasons for underscoring the “lead” in Exalead which, among other awards, was an LT-Innovate prize winner in 2012.

Semantic Vision
Exalead was founded in 2000 by two young computer scientists who wanted to build no less than the best and most comprehensive search engine from scratch. They heard the call of semantic technology before many others and decided to develop their own technology over the long term, ending up with a Semantic Factory that automates the whole process of aligning and enriching heterogeneous mixes of data. They also foresaw the need for searches over multimedia content, from video to voice, which has meant considerable investment into research as well as development.

The business vision paid off: in 2010 they were acquired by Dassault Systèmes (now known as 3DS) for around €150M. Their task was to round off the 3D design software company’s product lifetime management portfolio with a powerful search engine for the huge documentation and parts databases that underpin very large engineering projects such as airliners or nuclear power plants. But this move also gave Exalead access to 3DS’ database of over 115,000 customers.

Exalead now has a staff of 150 and a number of products and solutions that enable other industries such as healthcare, defence and finance or organisations such as contact centres to embed powerful search capabilities into very large data silos and then link them altogether to discover new insights.

The Exalead Ecosystem
As The Rude Baguette has shown in an excellent recent news story (to which this post is deeply indebted), Exalead has acted as a major incubator for twenty or so successful Paris-area start-ups. They have all been launched by former Exalead engineers in the past decade. Some of these businesses operate in the video or related search space but most of these younger companies draw in some way on Exalead’s strong culture of semantic technology skills. They obviously knew how to pick ambitious software engineers!

Today, Exalead is sufficiently resourced to continue with its own applied research agenda in a number of fields under the leadership of Chief Science Officer Gregory Grefenstette, who joined the company in 2008 after a distinguished career as an NLP researcher in the US and Europe. Exalead contributes to the open source community, develops innovative solutions to outstanding search problems, and above all has provided much of the inspiration and expertise for the €200M Quaero.

With its Latin for “I search” name, the largely Franco-German Quaero seems very much an emanation of Exalead’s original vision of seeing the universe of content through the eye of a search engineer. Its five areas of focus are personalized, multi-device distribution of video; better targeting for advertising; and multimedia search for the Web, all using the Exalead search engine.

Some of these application fields will presumably see products, apps or solutions emerging from the Quaero consortium – especially as the R&D phase is due to end in December this year. This means that the technology development process can begin. It’s worth noting that the project’s Voxalead application has already won three awards including a META 2nd Prize for outstanding audio visual search and transcription software and services in 2011. So it will be interesting to see which Quaero outcomes Exalead itself will productise in the next few years.

04 November 2013

New Opportunities for Virtual Assistants in European Customer Service

The IT consulting firm Gartner recently predicted  that Virtual Personal Assistant (VPA) usage in retailing contexts will grow more quickly in 2017 and 2018 than iPad usage did in 2010 and 2011. VPA technology can use a variety of channels – sometimes through text chat, but also using conversational interfaces on a smartphone – to connect customers with services that benefit from smart digital efficiencies.

Gartner has even claimed that in 2014 the number of speech recognition applications running on deep neural network algorithms (as touted by Google, Microsoft, Nuance and others today) will double, though most of them will probably handle personal information management tasks for individuals rather than customer services. So the question is: how can language and speech technology meet the specific challenges facing the customer-service market with respect to this evolution towards VPAs?

An interesting recent survey  by UK-based Creative Virtual, a CEM virtual assistant supplier, looked at the state of play and expectations for the immediate future in EMEA (Europe, the Middle East and Africa) countries and in the US for customer support challenges. The results are revealing.

Over a million inquiries a year
In the EMEA, 55% of those questioned would like to resolve customer inquiries faster. 40% want to reduce call volumes to live chat agents. And another 40% - over a third - want to step up the use of self-service channels. This presumably means installing the kind of technologies – search, natural language processing, and smart knowledge management – that can streamline the whole engagement process for customers.

To get a feel for the volume of calls or inquiries involved in customer service, and hence of the pressure on this service, 19% of the respondents referred to over one million inquires a year (close to 3,000 a day) in their businesses. Interestingly there appears to be a slight decrease in inquires to call centres and via email, presumably because more people are already using web channels such as social media or chat channels (a 45% increase). And of course customer service over mobile channels grew by 42%.

More self-service, less contact centre
What this means for businesses offering these CEM services is that increases in social, mobile and live chat suggest the rise of live/real time services over the web. This translates into an obvious opportunity for technologies such as virtual assistants, using text but tomorrow perhaps speech as an initial user search interface, or to engage directly with virtual avatars on service sites.

According to the survey, there has been faster “new channel” adoption in the EMEA, with 61% citing social media, and 55% seeing a rise in live chat. Unsurprisingly 62% of the respondents said they would be planning on developing social media channels in the coming years. This would presumably involve the use of smart technology to find and sort social responses to customer problems and personalise them to a given customer.

As for the tools they plan to roll out in the next 12 months, 23% mentioned virtual assistants, 21% online communities, and 16% cited forums. In the EMEA, 77% of those questioned already use FAQs, 69% feedback forms and 52% knowledge bases as primary service tools. This suggests that the emerging interfaces – VAs and speech – could both provide a faster, smoother front end to such inquires.
In all cases, these emerging trends in service management point to an increased use of conversation or a more “natural” interface whereby the richness of natural language will simplify life for human users, but require more advanced forms of language intelligence for tool and system suppliers.

Make sure your PVA is multilingual 
When it comes to PVAs, in the EMEA 70% of those questioned already use/plan to use virtual agents on their home page/customer service, 40% in the call centres, followed by 30% on live chat and on smartphones. In terms, of budget, respondents using PVAs said they devoted 1 to 10% of their budget to the solution. The next challenge is to make sure all these PVAs are as multilingual as is necessary in a single European marketplace.

Overall, then, a highly positive outlook for the numerous European VPA companies, from Artificial Solutions and Inbenta to Sherpa  - and even Creative Virtual itself.