Highlights from our semantic analysis of the language and word choices used by President Obama and Governor Romney in the first debate:
This week’s presidential debate is being analyzed across the web on a number of fronts, from a factual analysis of what was said, to the number of tweets it prompted. Instead, we used our Cogito semantic engine to analyze the transcript of the debate through a semantic and linguistic lens.
Cogito extracted the responses by question, breaking sentences down to their granular detail. This analysis allows us to look at the individual language elements to better understand what was said, as well as how the combined effect of word choice, sentence structure and sentence length might be interpreted by the audience.
Here is a sample of what we found:
Stay tuned for our analysis of the second debate on October 16.
The recent recognition of our customer, Telecom Italia, as the top corporate website in Europe gives us the chance to talk about the new semantic features we’ve recently incorporated at www.telecomitalia.it.
We have been working with Telecom Italia for some time now, applying our semantic technology to their customer service portal to quickly and precisely support customer requests. On its main portal, we apply additional semantic features to enhance information retrieval and provide users with a direct route to the information they’re seeking. These include:
The process combines the strengths of two of our products—Cogito Search Explore Engine (SEE), which accesses and analyzes all of Telecom Italia’s knowledge base, and Cogito Categorizer, used for categorizing large amounts of documents and content—resulting in high performance, optimized search results, and faster than before.
The advantage of Cogito is in its ability to distinguish between words and intention through an understanding of the meaning of words in context. In the form of a search portal, this translates to retrieving the precise information a user is searching for. The development of a custom taxonomy sorts documents automatically by interpreting their content, enabling users to concentrate on their area of interest, all without slowing the search process.
The world of search is continuously evolving, and today’s users expect not only immediate results, but new ways to view and explore them. We’re proud to see a long time customer recognized for a good website, because it underscores the value they place on information, and getting it to their users as quickly and easily as possible. That’s definitely a priority that we can relate to!
This year marked our fifth Semantic Technology Conference at the original San Francisco event. Coming to San Francisco each year is something I always look forward to, not just for the conference, but for the change of scenery and for the unique beauty and vibe that is San Francisco (and we even had a couple of sunny days this year).
For me, some highlights were the obvious and increasingly more visible merging of the world of data with the world of unstructured information and the strategic role the broadly defined semantic technologies play in it, and the growing number of implementations counting on a deep understanding of text . I also really enjoyed the workshop about ontologies, even if it got over my head pretty quickly.
I did feel like some of the program was a bit of a step back into academia, but maybe that is because there were less companies attending this year. I also noticed that most of the case studies featured are pilot projects or similar, while the real implementations are around natural language processing. I think this is because companies are more willing to share these kinds of initiatives than information about projects that really create a competitive advantage.
On the long flight back to Italy, I couldn’t help but think back to Semtechs past. I couldn’t put my finger on it at first, but I eventually realized that I missed that feeling of exclusivity, that this was the single event that everyone in the industry wouldn’t miss. It wasn’t for lack of people at this year’s event—there seemed to be enough attendees—but probably attributable to the fact that the conference, like the industry, has grown, and there are a lot more options and venues for taking part in the Semtech community. We’ve had the pleasure of speaking at a couple of the other newer venues, but I must admit that I miss the idea of feeling like, when we can all be everywhere all the time on the web, there is one central hub, one place where we all show up in person.
But, let’s face it, the application and recognition of semantics is growing—and that’s really good—and necessary—for our industry. The show program is a testament to companies in the same sector who are willing to collaborate for the good of the market. And the traditional data guys are also paying more attention to semantics, thanks to the ‘big data’ buzz. I think (and hope) that next year, we’ll see an even greater emphasis on the ‘Biz’ side of the conference (ROI, business driven applications of semantics, etc.), driven by some of these very things. And, I am sure I will be able to add some of our own stories next year!
Now that Google has released the first results of its new approach that is the semantic Knowledge Graph, we can finally evaluate its usefulness. (On a sidenote, I thought it was odd that KG was only available to registered users in the days immediately following its release. Maybe this was intended to give a slight advantage to those who provide Google with useful information for free every day?)
Now that I’ve had a chance to play with it a bit, I can say that this new functionality is less important for what it is—an intelligent display of information in their Freebase, enriched with elements extracted from the most visited search pages for a given query—than for what it represents: The first official step into the world of Google Semantics.
The Knowledge Graph concept is nice, and it can be a time saver in some situations. For example, when searching for a person or location, often you are indeed looking for the specific type of data that could be displayed by the KG.
But it’s nothing revolutionary. Others (including ourselves), made this years ago, although not on the same web scale for obvious reasons (cost, volume), while Google was arguing that there was no intelligent life beyond the keyword.
It is interesting to note that much of the information present in Freebase is extracted from Wikipedia (more or less automatically) and therefore, Google competitors can easily implement the same functionality.
For me, the real news is this: Semantic technology has finally emerged from its niche, becoming a technology that, in just a few years, will be pervasive. And you can’t avoid it. It’s what we’ve been saying, and doing, for more than 10 years now.
I am writing this post as my train pulls out of the Santa Lucia station in Venice, where I’ve had the pleasure of spending the past two days as part of the Digital Economy Forum (#DEF2012). Sponsored by the US Embassy in Italy, the DEF organized a series of events, all geared toward a more robust digital ecosystem in Italy. This evening, as Venice fades into the background, I wanted to share my thoughts on what has been an interesting couple of days and a very well organized, timely conference (a special thanks to Sandy Polu):
Italy and the digital world, part 1: Despite what people may believe, Italy is home to a large pool of quality software engineers and entrepreneurs. While the program was filled with interesting new companies who have worked hard to raise capital in hopes of becoming the next Glancee (an Italian company bought by Facebook), I expected to see a greater representation of the programmers and hackers among the “entrepreneurs to be.” At similar events, this group usually outweighs the MBA types, and, as a former programmer-turned-MBA-type, I’d like to see more Italian coders making their voices and ideas heard at these meetings.
Italy and the digital world, part 2: Despite Italy’s vibrant tech industry, I’m afraid that real understanding and support for our industry still lags behind at the government level. The speech of Minister of Economic Development (and former banker) Passera was more suitable for the general electorate (promises of solving the digital divide,open government data and updated websites), than for the crowd of Italian digital entrepreneurs who want inspiration, and concrete signs that those at the top understand our new world. While I appreciate the government’s acknowledgement, I had hoped for more signs that, like the Obama administration is constantly trying to do, our government is perceptive to the needs of this important and growing sector. For me, this was a missed opportunity.
The Speakers: Speakers included a nice mix of Italian and international companies, and the big guys were there too. Google spoke first, but instead of showing this international audience of entrepreneurs, investors and startuppers why they should see Google as a partner (which I would expect and hope to see given the discussions about privacy) and not an enemy, Google spent a lot of time explaining and promoting Google+. I know the days of “don’t be evil” are gone, and I’m not in a position to criticize their strategy, but as an old fan of the company, their transformation into the next IBM is impossible to ignore.
Instead, Facebook did a great job of showcasing companies who have made the most of the assets Facebook makes available, and called for revolutionaries (“Che Guevaras”) to join their movement. I hope this will translate to more and different applications rather than time-wasting games…
Among the most interesting companies and products who took the stage, I definitely suggest following Applix and Vertical Response. These companies offer a potential innovation boost for small- and medium-sized enterprises by providing them with innovative marketing tools and apps to help put them at par with bigger companies. My favorite presentations instead were those of Andraz Tori of Zemanta (they have a great product; try it if you’re a blogger) and Eleonora Viviani of Stereomood. As a music fan, I’ve spent years creating the perfect mixtape for every situation, so I understand the solution they’re striving for. From now on, their “Sunday morning” selection will be at the top of my weekend rotation.
But until then, I’ll go with this for the next two hours.
As a kid growing up in Milan I was obsessed with two things: sports and the future. If I wasn’t in the local park playing soccer, I was lost inside an issue of the futuristic sci-fi comic El Eternauta (upon reflection, perhaps a bit dark for a kid!). In those days, Italian television was a mere two channels and little more than Eastern European cartoons or variety shows on Saturdays, and sports events were broadcast several hours after the live match.
While other kids probably dreamed about having superpowers, my dreams were about a future that was more grounded in possibility than fantasy, about inventions that, with the right combination of genius and determination, were possible. Having my own personal television, one small enough to take anywhere and where I could watch anything I wanted, anytime, was something I knew would come eventually. And while the mini TV (remember this, or this?) was an encouraging sign, the technology was still many years away.
Fast forward to today… I’m watching Mad Men on my iPad and in my still pragmatic dreams, it’s not a leap to think that twenty years from now, we will be living in a very self-curated world. I imagine a world customized for me, my every activity synched and automated based on my personal calendar and a pre-defined set of parameters. A lunch meeting would automatically trigger a series of actions based on the meeting attendees, our location and our preferences. Reservations would be made, train tickets or flights purchased automatically and without any additional input from me.
I see hints of this “reality” today, just as I could imagine that one day, I’d be watching Serie A on the train ride home. While there are many technological and behavioral (not to mention security and privacy-related) boundaries to be crossed in the meantime, semantic technology is an important piece of this puzzle.
Data/information does not exist in a vacuum, and it is becoming easier to interconnect it with systems and technologies to design products that define a user experience increasingly similar to my idea of a “realistic future”.
Semantic technology excels in areas that present the biggest challenges for our information and data driven society by resolving context and connection within and between information. Semantics is increasingly being used to connect information silos and resolve disparities in data (particularly useful in the cases of acquisitions or consolidations), resulting in greater visibility of information assets and intelligence that can be translated to product innovation.
In this way, semantics will bridge gaps, contributing to the more connected, accessible and intuitive world that I imagine.
The financial sector consolidations over the past decade left many firms with volumes of knowledge trapped in legacy systems—but only a surface-level understanding of the valuable data they acquired.
Understanding how to best leverage the fruits of their consolidations—data—is where many are now focusing their attentions and IT budgets. An event I attended last month, “Demystifying Financial Services Semantics: The Business Value of Data and Semantics” hit the point home: Big Data has arrived, and while not without challenges, it presents a world of opportunity.
Financial institutions are looking to linguistics and semantics as the best option for managing and taking advantage of their unstructured data, using it to better understand customers and competitors, to identify impactful market trends or simply to automate the process of answering common customer questions.
As one Chief Data Officer put it:“We are the stewards of one of our firms’ most important assets, data, and we have been charged with bringing meaning to the data. I believe semantics offers a consistent, long-term capability and change in how data will be managed.”
The ability to understand words in context is where semantics has proven to excel above all other technologies—here’s how:
We are seeing customers take advantage of semantic technology by not only analyzing customer data to understand and benefit from customer feedback, but also through automated, virtual assistance.
Big Data presents a challenge for many industries and financial institutions are stepping up to that challenge, whether through the development of the Financial Industry Business Ontology (FIBO), or through Big Data projects geared toward implementing a framework to support the Dodd-Frank Act and data transparency.
Big Data is here to stay, and it is clear that having the ability to understand words in context using a semantic intelligence platform is critical for driving the new revolution.
Google’s interview in The Wall Street Journal last week stirred up lots of discussion, and one of Italy’s main dailies called us up that afternoon to get our take on what Google’s overhaul of search will really mean.
First, it’s a question of content. The proclamation that Google is working on a next-generation search engine that incorporates semantic search is more of a shortcut than a solution. What Google is still missing, at least as far as we know, is the ability to extract all data and information from any web page (and not just the data taken from a reference database like FreeBase, a company acquired by Google who created a database of all the data/entities in Wikipedia). Implementing a semantic search engine requires additional steps. For example, semantic search should understand the multiple meanings a word can have (which depends largely on context), not to mention the need to adapt to different languages and cultures. In addition it should also be improved to address the “doubts” — testing the system with real and simulated questions to identify its weaknesses. This is only the beginning—real semantic search technology must address all of these areas.
And finally, it’s the how of the announcement, really, that is interesting. This is the first time I remember Google ‘announcing’ something in advance, rather than on the heels of something it’s already done. Maybe it’s just a change in PR strategy, or does it signal a bigger internal (and more corporate) shift away from its roots?
The value of unstructured information to an organization’s business intelligence process goes well beyond sentiment analysis. Unstructured information, if properly utilized, can fill in the gaps between the polarity of “Like” or “Dislike” and provide the why (why you choose them), how (how you feel) and where (where you stand) that every company should want to know.
Today, I’d like to share some real-world examples of customers who are using semantic technology and robust tools for natural language understanding to derive real business intelligence (and not just sentiment) from unstructured information.
Knowing your ‘friends’. Every company with a Facebook page and thousands of ‘friends’ to match has a wealth of market research at its fingertips—whether they actually convert to customers or not, knowing more about your ‘friends’ can only help your business and product strategy.
Information shared on social networks—preferences, tastes, desires—can be translated into valuable business intelligence when integrated into your existing data—demographics, sales data and other statistics—adding great depth and insight to your customer model. As one of our customers is currently experiencing, implementing a good semantic product can help you not only understand your customers, but also how the values you are trying to communicate through your brand are perceived, and what customers think of you vs. the competition. The advantages are significant: up-to-the-minute insight that is significantly less expensive compared to traditional BI or market research.
From questions to answers. When companies implement self help solutions, their main focus is on the savings achieved by deflecting calls from their customer service center. We calculated a $2 million/month savings for one customer—a major manufacturer of handhelds—simply by implementing a self help application on its devices. What the numbers don’t tell us—and semantic technology does—is that the ROI goes much deeper.
Using semantic technology to analyze the inquiries customers made through the self help application gave us valuable insight that allowed our customer to:
This information—all from existing data—adds more dimension to customer information, and provides not just savings, but intelligence that can impact every area, from sales and marketing to product development and more.