Daniel Tunkelang: Co-Founder of Endeca Interviewed

February 9, 2009

As other search conferences gasp for the fresh air of enervating speakers, Harry Collier’s Boston Search Engine Conference (more information is here) has landed another thought-leader speaker. Daniel Tunkelang is one of the founders of Endeca. After the implosion of Convera and the buys out of Fast Search and Verity, Endeca is one of the two flagship vendors of search, content processing, and information management systems recognized by most information technology professionals. Dr. Tunkelang writes an informative Web log The Noisy Channel here.


Dr. Daniel Tunkelang. Source: http://www.cs.cmu.edu/~quixote/dt.jpg

You can get a sense of Dr. Tunkelang’s views in this exclusive interview conducted by Stephen Arnold with the assistance of Harry Collier, Managing Director, Infonortics Ltd.. If you want to hear and meet Dr. Tunkelang, attend the Boston Search Engine Meeting, which is focused on search and information retrieval. The Boston Search Engine Meeting is the show you may want to consider attending. All beef, no filler.


The speakers, like Dr. Tunkelang, will challenge you to think about the nature of information and the ways to deal with substantive issues, not antimacassars slapped on a problem. We interviewed Mr. Tunkelang on February 5, 2009. The full text of this interview appears below.

Tell us a bit about yourself and about Endeca.

I’m the Chief Scientist and a co-founder of Endeca, a leading enterprise search vendor. We are the largest organically grown company in our space (no preservatives or acquisitions!), and we have been recognized by industry analysts as a market and technology leader. Our hundreds of clients include household names in retail (Wal*Mart, Home Depot); manufacturing and distribution (Boeing, IBM); media and publishing (LexisNexis, World Book), financial services (ABN AMRO, Bank of America), and government (Defense Intelligence Agency, National Cancer Institute).

My own background: I was an undergraduate at MIT, double majoring in math and computer science, and I completed a PhD at CMU, where I worked on information visualization. Before joining Endeca’s founding team, I worked at the IBM T. J. Watson Research Center and AT&T Bell Labs.

What differentiates Endeca from the field of search and content processing vendors?

In web search, we type a query in a search box and expect to find the information we need in the top handful of results. In enterprise search, this approach too often breaks down. There are a variety of reasons for this breakdown, but the main one is that enterprise information needs are less amenable to the “wisdom of crowds” approach at the heart of PageRank and related approaches used for web search. As a consequence, we must get away from treating the search engine as a mind reader, and instead promote bi-directional communication so that users can effectively articulate their information needs and the system can satisfy them. The approach is known in the academic literature as human computer information retrieval (HCIR).

Endeca implements an HCIR approach by combining a set-oriented retrieval with user interaction to create an interactive dialogue, offering next steps or refinements to help guide users to the results most relevant for their unique needs. An Endeca-powered application responds to a query with not just relevant results, but with an overview of the user’s current context and an organized set of options for incremental exploration.

What do you see as the three major challenges facing search and content processing in 2009 and beyond?

There are so many challenges! But let me pick my top three:

Social Search. While the word “social” is overused as a buzzword, it is true that content is becoming increasingly social in nature, both on the consumer web and in the enterprise. In particular, there is much appeal in the idea that people will tag content within the enterprise and benefit from each other’s tagging. The reality of social search, however, has not lived up to the vision. In order for social search to succeed, enterprise workers need to supply their proprietary knowledge in a process that is not only as painless as possible, but demonstrates the return on investment. We believe that our work at Endeca, on bootstrapping knowledge bases, can help bring about effective social search in the enterprise.

Federation.  As much as an enterprise may value its internal content, much of the content that its workers need resides outside the enterprise. An effective enterprise search tool needs to facilitate users’ access to all of these content sources while preserving value and context of each. But federation raises its own challenges, since every repository offers different levels of access to its contents. For federation to succeed, information repositories will need to offer more meaningful access than returning the top few results for a search query.

Search is not a zero-sum game. Web search engines in general–and Google in particular–have promoted a view of search that is heavily adversarial, thus encouraging a multi-billion dollar industry of companies and consultants trying to manipulate result ranking. This arms race between search engines and SEO consultants is an incredible waste of energy for both sides, and distracts us from building better technology to help people find information.

With the rapid change in the business climate, how will the increasing financial pressure on information technology affect search and content processing?

There’s no question that information technology purchase decisions will face stricter scrutiny. But, to quote Rahm Emmanuel, “Never let a serious crisis go to waste…it’s an opportunity to do things you couldn’t do before.” Stricter scrutiny is a good thing; it means that search technology will be held accountable for the value it delivers to the enterprise. There will, no doubt, be an increasing pressure to cut costs, from price pressure on vendor to substituting automated techniques for human labor. But that is how it should be: vendors have to justify their value proposition. The difference in today’s climate is that the spotlight shines more intensely on this process.

Search / content processing systems have been integrated into such diverse functions as business intelligence and customer support. Do you see search / content processing becoming increasingly integrated into enterprise applications? If yes, how will this shift affect the companies providing stand alone search / content processing solutions? If no, what do you see the role of standalone search / content processing applications becoming?

Better search is a requirement for many enterprise applications–not just BI and Call Centers, but also e-commerce, product lifecycle management, CRM, and content management.  The level of search in these applications is only going to increase, and at some point it just isn’t possible for workers to productively use information without access to effective search tools.

For stand-alone vendors like Endeca, interoperability is key. At Endeca, we are continually expanding our connectivity to enterprise systems: more connectors, leveraging data services, etc.  We are also innovating in the area of building configurable applications, which let businesses quickly deploy the right set features for their users.  Our diverse customer base has driven us to support the diversity of their information needs, e.g., customer support representatives have very different requirements from those of online shoppers. Most importantly, everyone benefits from tools that offer an opportunity to meaningfully interact with information, rather than being subjected to a big list of results that they can only page through.

Microsoft acquired Fast Search & Transfer. SAS acquired Teragram. Autonomy acquired Interwoven and Zantaz. In your opinion, will this consolidation create opportunities or shut doors. What options are available to vendors / researchers in this merger-filled environment?

Yes!  Each acquisition changes the dynamics in the market, both creating opportunities and shutting doors at the same time.  For SharePoint customers who want to keep the number of vendors they work with to a minimum, the acquisition of FAST gives them a better starting point over Microsoft Search Server.  For FAST customers who aren’t using SharePoint, I can only speculate as to what is in store for them.

For other vendors in the marketplace, the options are:

  • Get aligned with (or acquired by) one of the big vendors and get more tightly tied into a platform stack like FAST;
  • Carve out a position in a specific segment, like we’re seeing with Autonomy and e-Discovery, or
  • Be agnostic, and serve a number of different platforms and users like Endeca or Google do.  In this group, you’ll see some cases where functionality is king, and some cases where pricing is more important, but there will be plenty of opportunities here to thrive.

Multi core processors provide significant performance boosts. But search / content processing often faces bottlenecks and latency in indexing and query processing. What’s your view on the performance of your system or systems with which you are familiar? Is performance a non issue?

Performance is absolutely a consideration, even for systems that make efficient use of hardware resources. And it’s not just about using CPU for run-time query processing: the increasing size of data collections has pushed on memory requirements; data enrichment increases the expectations and resource requirements for indexing; and richer capabilities for query refinement and data visualization present their own performance demands.

Multicore computing is the new shape of Moore’s Law: this is a fundamental consequence of the need to manage power consumption on today’s processors, which contain billions of transistors. Hence, older search systems that were not designed to exploit data parallelism during query evaluation will not scale up as hardware advances.

While tasks like content extraction, enrichment, and indexing lend themselves well to today’s distributed computing approaches, the query side of the problem is more difficult–especially in modern interfaces that incorporate faceted search, group-bys, joins, numeric aggregations, et cetera. Much of the research literature on query parallelism from the database community addresses structured, relational data, and most parallel database work has targeted distributed memory models, so existing techniques must be adapted to handle the problems of search.

Google has disrupted certain enterprise search markets with its appliance solution. The Google brand creates the idea in the minds of some procurement teams and purchasing agents that Google is the only or preferred search solution. What can a vendor do to adapt to this Google effect? Is Google a significant player in enterprise search, or is Google a minor player?

I think it is a mistake for the higher-end search vendors to dismiss Google as a minor player in the enterprise. Google’s appliance solution may be functionally deficient, but Google’s brand is formidable, as is its position of the appliance as a simple, low-cost solution. Moreover, if buyers do not understand the differences among vendor offerings, they may well be inclined to decide based on the price tag–particularly in a cost-conscious economy. It is thus more incumbent than ever on vendors to be open about what their technology can do, as well as to build a credible case for buyers to compare total cost of ownership.

Mobile search is emerging as an important branch of search / content processing. Mobile search, however, imposes some limitations on presentation and query submission. What are your views of mobile search’s impact on more traditional enterprise search / content processing?

A number of folks have noted that the design constraints of the iPhone (and of mobile devices in general) lead to an improved user experience, since site designers do a better job of focusing on the information that users will find relevant. I’m delighted to see designers striving to improve the signal-to-noise ratio in information seeking applications.

Still, I think we can take the idea much further. More efficient or ergonomic use of real estate boils down to stripping extraneous content–a good idea, but hardly novel, and making sites vertically oriented (i.e., no horizontal scrolling) is still a cosmetic change. The more interesting question is how to determine what information is best to present in the limited space–-that is the key to optimizing interaction. Indeed, many of the questions raised by small screens also apply to other interfaces, such as voice. Ultimately, we need to reconsider the extreme inefficiency of ranked lists, compared to summarization-oriented approaches. Certainly the mobile space opens great opportunities for someone to get this right on the web.

Semantic technology can make point and click interfaces more useful. What other uses of semantic technology do you see gaining significance in 2009? What semantic considerations do you bring to your product and research activities?

Semantic search means different things to different people, but broadly falls into two categories: Using linguistic and statistical approaches to derive meaning from unstructured text, using semantic web approaches to represent meaning in content and query structure. Endeca embraces both of these aspects of semantic search.

From early on, we have developed an extensible framework for enriching content through linguistic and statistical information extraction. We have developed some groundbreaking tools ourselves, but have achieved even better results by combining other vendor’s document analysis tools with our unique ability to improve their results through corpus analysis.

The growing prevalence of structured data (e.g., RDF) with well-formed ontologies (e.g., OWL) is very valuable to Endeca, since our flexible data model is ideal for incorporating heterogeneous, semi-structured content. We have done this in major applications for the financial industry, media/publishing, and the federal government.

It is also important that semantic search is not just about the data. In the popular conception of semantic search, the computer is wholly responsible derives meaning from the unstructured input. Endeca’s philosophy, as per the HCIR vision, is that humans determine meaning, and that our job is to give them clues using all of the structure we can provide.

Where can I find more information about your products, services, and research?

Endeca’s web site is http://endeca.com/. I also encourage you to read my blog, The Noisy Channel (http://thenoisychannel.com/), where I share my ideas (as do a number of other people!) on improving the way that people interact with information.

Stephen Arnold, February 9, 2009


One Response to “Daniel Tunkelang: Co-Founder of Endeca Interviewed”

  1. Previews of Upcoming Industry Search Conferences | The Noisy Channel on February 15th, 2009 11:19 am

    […] Beyond Search blog,  has been publishing interviews with some of the speakers. You can find mine here. I’m partial to this format, since I find text a more efficient medium for this sort of […]

  • Archives

  • Recent Posts

  • Meta