|Sentius Corp. is the developer of RichLink, a patented database linking
and embedding technology that automatically adds content to Web sites.
RichLink embeds layers of relevant content that pops up at the reader's
request in a window called a "Knowledge Burst." The company recently launched
its e-publishing suite, which is designed to increase advertising and licensing
revenues for publishers. I talked with Marc Bookman, Sentius Corp.'s CEO
and founder, about the company's technology, markets, and plans for the
Q Content enrichment or enhancement on Web
sites is usually a manual process done by editors who code or write annotations
and decide upon and provide links. Let's talk about the concept of "automated
content enrichment," for readers who might not be familiar with this technology.
A The goal of automatic content enrichment
is to enable publishers to rapidly layer additional content and functionality
into their documents in a highly scalable manner. The enrichment focuses
on helping readers to understand, decide, and act more effectively on that
information. One example of content enrichment is enabling English documents
with foreign-language assistance to make non-native readers proficient.
Another example would be to enable a customer-support person to read a
technical document and answer customer inquiries more quickly. Ultimately,
Web sites should be able to publish documents to a worldwide audience with
the best references based on the language, proficiency, and relationship
that they have with the end-user.
The process starts with an analysis of the HTML or XML document to determine
the important parts of the page, and then automates and executes the rules
for linking. Following this structural analysis, we do a linguistic analysis
of words, sentences,and terms. Though it may sound simple, it's quite complex.
We've invested tens of thousands of man-hours in the technology to make
Q And is this proprietary technology that
was entirely developed by Sentius?
A We built everything from scratch: all
the components, the architecture, and our own English-language parser.
Recently, we added Inxight's LinguistX package into our system. It's great
for analyzing inflections or morphology and doing things like stemming.
Q What kind of content is included in your
pop-up Knowledge Bursts besides definitions and descriptions?
A We work closely with our customers to
help them crystallize their ideas of what to incorporate into their Knowledge
Bursts. That can mean company information, term definitions, marketing
messages, and so on. Lately, we've been hearing more requests from our
customers and prospects for RichLink-enabled product information that would
show the most relevant product info when a reader clicks on a product name.
We think that's going to be a big area. We also see great interest in Web
sites being able to link their key terms to the latest news items on their
site. There are images in some of our content sources and many of our customers
intend to include movies down the road. When the tools improve for audio
and visual search access, we will be able to link more effectively to multimedia
Q Let's return to our discussion of the
term "automated." It would seem that a great deal of people involvement
would be required in addition to the automated process, especially in the
initial setup process.
A Correct, though our foreign-language application
is at the stage where it's pretty easy for us to set up an e-globalization
solution for someone. Even if customers have their own glossaries it can
be done fairly quickly. For other RichLink applications on sites that are
set up well with a decent search engine and taxonomy, we can basically
implement an out-of-the-box solution. But, if we really want to go in and
evaluate the content and the layers of information, it does take some planning
and solid editorial work. Once the system is implemented, however, the
process is thereafter totally automated.
Q Tell us about the background of starting
the company and developing RichLink. Was it originally developed just to
deal with language translations?
A I was in a sea of Japanese-language documents
and e-mail when I was living in Japan and working for Sony, and that's
when the need became clear to me. When I started the company, I wasn't
sure exactly which application to pursue first. When I showed prototypes
to people, I heard, "I wish I had had this for my chemistry textbook,"
or "I wish I could read technical documents this way." I chose foreign-language
annotation as the first application since the need for it is so clear.
We immediately moved on knowing that foreign-language applications were
just the tip of the iceberg.
Q We all like information to be available
at the point of need—you call it "at the point of impact." Visually what
RichLink provides is different: Clicking doesn't send the user off to another
page or site or browser window; the information pops up in a box. How does
your automated technology actually differ from other competitors' technologies
(such as Atomica), both technically and from the users' experience? You've
patented your technologies, but why couldn't competitors just copy the
concept of a pop-up box?
A Our intellectual property relates to taking
a textual document with no structure, no links, no tagging, and ending
up with a document with enriched links that guide the user as to what to
click on. And when the user clicks on those anchor points in the document,
pop-up information is provided. It's the combination of all those things
that we patented. We provide direct linkages between a term on a page and
content in a database or in multiple database sources. We give the user
database content within context.
We thought this entire process through some time ago, and there have
been a few companies that have tried to employ a similar process. Flyswat,
the company that NBCi purchased, is one. There are some other companiesnow
releasing technology that driveslinks onto Web pages and documents via
the contents of databases or external reference sources.
What Atomica is trying to do is very interesting. The goal of aggregating
information pop-up portals based on words and phrases is a similarity between
our two companies. Right now they seem to be focused on corporate intranet
applications. They have a desktop client that a user installs, and then
when a user presses ALT and clicks on a term, a pop-up window appears.
It's a neat application and is actually an approach that we toyed with
many years ago, but we really wanted to get out of trying to sell a search
tool to end-users—either to the corporate market, which we felt would be
a difficult sell and require dealing with IT departments, or to consumers,
where the ceiling on how much they are willing to spend on a reference
work is pretty low. Our customer is a Web site or a publisher that is syndicating
or publishing content to the Web. We are trying to help them turn their
documents into richer portals of information.
Q In the databases
of information that you have available for licensing, I understand that
you have over 3 million terms in 15 industry areas. Who are the content
partners that provide these databases and what are some of the databases
that our readers might know?
Japan, we're working with Kenkyusha, a dictionary provider, and also with
Nova, a large translation software company, so we have top-of-the-line
Japanese reference information. For European languages, we're working with
Lernout and Hauspie, as well as some smaller vendors who are doing more
specialized European language dictionaries for medical and other scientific
fields. For each category, we try to identify who the top-branded source
is and work out a partnership. For some of our U.S. domestic content, we've
worked out relationships with Academic Press (The Dictionary of Science
and Technology), Merriam-Webster, Facts and Comparisons and Lexi-Comp (both
drug database providers), MedicineNet (see Figure 1),
as well as some general sources like American Heritage and Columbia Encyclopedia.
These sources serve as a base for our customers to work with, if they choose.
They can then add custom layers or work through us to add other content.
Q RichLink is available either as a hosted
service or for on-site installation, isn't it?
A Yes, we offer the software and a full
end-to-end hosted service, and also hybrids of these. For example, people
will use our software but have us manage their databases for them.
Q With such a range of possible installations,
is it possible to quote an average cost or a price range for this?
A Yes, the cost can start as low as $25,000
and range up to about $250,000, depending on what is licensed, how many
servers, and how many databases are licensed.
Q You seem to have a fairly large range
of target markets: content publishers, portals, corporate extranets, etc.
As a small company, how do you sell to and support this diverse group?
A Our focus today is clearly on three vertical
markets: medical and pharmaceutical, information technology, and the financial
services industry. Our strategy is to work with top-line publishers in
each space, licensing either our technology or services to them. We're
also interested in working with other solutions providers, either those
providing content management systems or content technology to these marketplaces.
We want to embed our solution within these other platforms. We have worked
with quite a few enterprise customers ourselves, but we think a more effective
approach for us is to work through these channel partners, which have larger
sales forces and already have relationships with enterprise customers.
Content companies that are making their databases available to affiliate
content sites on the Web, by working with us can offer the ability to embed
their content directly in the pages of their customers, and expect to greatly
increase the viewership of that content on their customers' pages. An example
of this is Facts and Comparisons, which saw the opportunity to complement
the drug look-up functionality of their data on their customers' sites
with direct access to their content at the word and phrase level of their
customers' sites using RichLink. (See image on page
A publisher like Reuters Health has a news feed that they deliver to
a large number of sites, and they want to expand the number of sites internationally
as well as provide value-added solutions to those sites. RichLink is helping
them build new audiences and build more functionality into their news.
Q What prompted this discussion today was
the recent announcement of the launch of your e-publishing suite, with
Word Burst for increasing ad revenue, and License Burst for embedding licensed
content and increasing syndication revenue. Let's focus for a minute on
the targeted ads popping up in these windows. As a user I find great dislike
for ads that pop up automatically—they are annoying and intrusive.In your
case though, the user has to opt to activate an ad by clicking on a word,
so I would guess users don't object and advertisers and publishers must
A That's right. We encourage customers not
to build in pure advertising as pop-ups, since people will just stop clicking.
When users are rewarded with high-value content, they can then be offered
other messages within that same window. This is not about in-your-face
advertising; it's about providing high-value information within context
to a user. We get the end-user at the point when they want to know more—and
for an advertiser to get a user at that point is a powerful way to get
your message across. Our point-of-impact slogan was inspired by the education
world, which talks about the "teachable moment." If you can talk to someone
at that teachable moment, your effectiveness is much higher.
Q Publishers and editors have long been
concerned about the lines of distinction between editorial content and
advertising. Moving to the Web, we've been even more observant and careful.
From what I've seen though, in your examples, it's very clear when a piece
of database information is presented and when the user is invited to click
on an advertisement for more information.
A Yes, it is. We're very active now in the
medical and pharmaceutical space where the sensitivity is very high on
this topic. We enable our customers to customize the pop-up window however
Q How does the License Burst system actually
work? Your press release mentioned that it "works seamlessly with any content
management system." What exactly is required?
A This answer actually applies to RichLink
in general and not just to License Burst. We will plug into almost any
system: homegrown CMS; a flat Apache Web server environment; or into a
more complex environment, such as the Documentum content management system.
A site running databases with great work flow and a taxonomy in place will
be able to do more with RichLink than a site than doesn't have those things.
For License Burst, we want to work with companies that are already selling
their content on the Web to other sites and affiliates. We want to enable
them to offer another version of their content, not just a search capability
but a RichLink capability to embed their content in another site. It's
up to the customer to figure out how to license it and charge for it. For
example, because Reuters Health has a version of their medical content
RichLink-enabled with Japanese-language annotations, they expect to be
able to double their content licensing in Japan.
Q So License Burst doesn't enable someone
to find an article and request to license it for another site—something
like Qpass, which allows one to purchase an article.
A Someone could certainly integrate Qpass
with RichLink, and use the RichLink capability to direct people to the
other content. We have people who are thinking about doing that. We do
have people exploring the use of RichLink to drive archive sales. It's
one of the next things we'll be focusing on.
are a privately held company, so I don't know how you've been doing in
these difficult market conditions. Has this been tough for you? You launch
a new company, get off to a good start with your technology, and then watch
potential customers scale back their software and technology purchases
as they struggle with their bottom lines.
A The most difficult times were actually
in January and February of this year. The sales pipeline in December of
last year was very promising, but by the end of February folks were disappearing
from companies. Projects were being canceled. There were layoffs, delays
in Web site plans—our customers were actually disassembling themselves.
We are now seeing interest pickup within the publishing space, so we've
put considerable effort into our e-publishing suite. We're offering publishers
new ways to monetize their content. In this kind of market, companies want
to be offered ways to increase their revenue, create new products, and
raise customer satisfaction. We're getting good traction and we're incredibly
busy right now.
Q In fact, a solution like RichLink couldreally
help companies that are struggling now to enable their communication and
their e-commerce to be more efficient and productive. Maybe you are in
the right place at the right time.
A It's not just the economy. I think there's
also been a natural progression of electronic media technologies. Web technology
is starting to stabilize and people are looking to build greater functionality
and improve customer satisfaction, just as happened earlier with CD-ROM
publishers. Web sites know their end-users better now and are looking for
ways to add real value.
Sentius Corp. is based in Palo Alto, California. For more information,
RichLink Word Burst can be seen in action on MedicineNet at http://www.medicinenet.com/richlink.html,
and RichLink e-globalization can be seen at http://www.reutershealth.com/frame2/eline_jp.html.
RichLink License Burst as being sold by Facts and Comparisons can be seen
Paula J. Hane is editor of NewsBreaks, contributing editor
of Information Today, a former reference librarian, and a
longtime online searcher. Her e-mail address is email@example.com.