By John-Scott Dixon
My five-year-old son recently asked me, "What is ground?" I responded with, "It is the surface of our planet." He said "No, ground." Puzzled, I pressed on with, "Well, when they make hamburger, they put steak into a grinder, then it is ground into hamburger meat." Frustrated, he said, "No Dad, ground." Finally, I replied with, "Are you talking about 'grounded,' like when kids get in trouble?" He said, "Yes, what is ground?" We were dealing with semantics.
The Semantic Web, or Web 3.0, is about meaning - understanding the context of a word or concept to offer relevant resources, thereby making it easier for people to find what they seek. In the example at left, my son was the user and I was the search engine, having trouble understanding the context of his query.
As the Web and its available resources continue to expand, it is becoming increasingly difficult to find what you want efficiently. It's like stocking the shelves of your local grocery store with nothing but generic brands and expanding it to the size of five city blocks. We need help making choices.
The Semantic Graph
When I search, the list of results can be daunting - a search on Google for "ground" returned over five million results. Over the last few years, we have also been able to expand our searches to images, podcasts and video as well as websites. With proper syntax, we can structure searches that give back information only - like how many cups in a pint? But for anything other than weights and measures, the search engines are often lacking.
For example, if you want to find the earned run average (ERA) for major league pitcher Randy Johnson for every season he's played, it would take a massive effort on Google. Search "Randy Johnson ERA" and Google will direct you to websites that might contain some of the information you are after, like ESPN.com, SI.com, the Arizona Diamondbacks' website or BaseballLibrary.com. Those websites will certainly contain data about Randy Johnson and his ERA, but you will have to do some more digging to find a list of his ERA from the beginning of his Major League Baseball career through last season. ecently, I spent some time with John Grace - CEO of ENTH, a next generation search engine (Enth.com). We performed the Randy Johnson search, and the result was an ordered list of his ERA from 1988 through 2007, rather than a list of websites that might, or might not have the information. ENTH is only making sense of Major League Baseball, the National Basketball Association and College Football for now, but their mission is to apply their technology to every database that exists. And there are other semantic search engines changing the way we retrieve information, well beyond the sports arena.
Each of us has a unique set of things that we value - people, places, experiences and objects we deem important. And although our interests are personal and unique, there are countless others who may share some of these interests, or have interests related in some way to our own. That creates common themes across categories, and people who share common interests in each of those categories. There may be relationships that we're unaware of that exist, like people interested in visiting Kauai who are also interested in vegetarian diets. This is what Nova Spivack, founder of Twine, calls a Semantic Graph. By using the Semantic Graph, a new brand of search engines and Web applications can make suggestions based on our unique combination of interests and the interests of people in our networks.
What Semantic Search Means to Your Business
So why aren't we seeing much in the way of semantic search from Google? A likely theory is that it would hurt their core business of advertising. Russell Glass, general manager and senior vice president of advertising for ZoomInfo says, "if their search results were perfect, who would click on the ads?" That's a valid point, and one that presents an opportunity for new search engines. The challenge is creating a business model where contextual advertising is less powerful.
ZoomInfo offers a $99 annual membership to perform people searches for sales and marketing personnel building prospecting lists. ENTH will charge businesses a premium to mash internal databases with external public databases. As for PowerSet, it is still unclear how they will monetize semantic search.
What can be done today and how do we prepare for tomorrow? Here are five suggestions:
1. Connect with the right people. Imagine your company has a product that takes advantage of mobile telephony for use in automobiles. And, one of your most promising conversations was with an executive at an automotive company who previously worked in the telecommunications industry. With technology like ZoomInfo, you might be able to build a list of executives who meet that profile. That hasn't been possible before. Available Today.
2. Make industry data available for public inquiry. Using the power of ENTH or a similar technology enables people to find relationships in your data that you might not have otherwise found. The benefit is the association of your brand as the source of helpful information. Available Today.
3. Educate semantic search engines by defining your commercial space to ontology engines like WordNet from Princeton University (WordNet.Princeton.edu). As an example, Fender Guitar Company may want to teach the semantic search engines that Fender manufactures acoustic and electric guitars, and within the electric guitar category is the Stratocaster, Telecaster, etc. The idea is when somebody searches for "Best Electric Guitar," the search engine might find a blog where a guitarist claims his Stratocaster is the best guitar ever made. A semantic search engine, like Powerset, would understand a reference to a Stratocaster as "best" is related to the search for "Best Electric Guitar." It would know the Stratocaster is an electric guitar and a user says it's the best. Available Today.
4. Create landing pages to welcome visitors who represent market segments most likely to visit your website. A common example is a competitive shopper - someone visiting the websites of your competitors and, more importantly, browsing specific categories, products or services. If you knew they were comparing your product or service to a specific competitor's, how could you greet them to be more persuasive in the sales process? Detect visitors that match those specific market segments when they arrive and display the corresponding landing page. Available Today.
5. When Twine launches this spring, companies will be able to create "twines" around commercial topics. This puts them at the heart of conversations about their commercial space. It will also be important to establish a Semantic Graph for your company. I am seeing Semantic Graphs as a form of digital fingerprint. It might be the new way to express your brand. So people with certain attributes are attracted to companies that cater to those attributes. Available Summer 2008.
More on Semantics
ZoomInfo (ZoomInfo.com)
This search engine is very business focused, allowing us to drill down on companies, people and jobs through Q&A-based searching. The results are better organized business information for sales and marketing purposes. As an example, you can search "Technology + Arizona" and get over 100 profiles of technology companies located in Arizona. A link to each of their websites is just a part of the information provided.
Powerset (Powerset.com)
The Powerset service will launch in a few months. It is able to extract meaning by reading documents like website pages, sentence by sentence. So, if you're interested in finding a list of swimsuit manufacturers, Powerset would have already examined every website out there that discusses swimsuits, bathing suits, bikinis, etc. It also has combined those words with others like manufacturer, maker, designer, etc. So, the result is a comprehensive list of swimsuit manufacturers. While Google will return websites that list swimsuit manufacturers, the results are intermingled with less meaningful results as well.
Semanticator (Semanticator.com)
ThougtLava's Semanticator enables marketers to translate their target markets or market segments into semantic personas which can be detected moments before arrival. Each persona or profile represents a combination of attributes like geographic location, operating system, day of the week, search keywords, targeted websites visited, etc. By detecting market segments before arrival, marketers are able to introduce visitors matching a particular persona to their products and services in a more meaningful way.
Twine (Twine.com)
This is a product of Radar Networks designed to help us discover our world in ways that we'll find interesting. As we organize and share interests, we'll discover new information within those interests and people who are equally as passionate. As an example, I'm interested in Arizona travel. So, I might post references to articles about places I'd like to visit, like the Grand Canyon. Others on Twine may find me because of my Grand Canyon post - and we may choose to connect in order to share additional Arizona travel information directly. As my relationships grow, I can immediately alert my Arizona travel group to any new information I've found. As the library of content around Arizona travel grows, newcomers will find it when searching.
About the Author: John-Scott Dixon is the Founder of ThoughtLava. See a demonstration of their patent-pending technology at www.Semanticator.com.