Enterprise Search Sourcebook 2008

EDITOR’S NOTE Eye Prize on the t o most people, search seems easier than ever before. One needs only to key a word or two into a single box on an uncluttered screen and hit Return in order to receive a plethora of results. This simple input, massive output principle adequately serves many searchers’ needs. Yet to those for whom a day’s work involves seeking information—which can make or break a major deal—the stakes are higher. The techniques, and often tools and tactics, employed take on a more complex form in kind. Without doubt, the value of search continues to rise for information seekers and solution providers alike. It is not an illusion that search has become easier for end users; usability sits at the top of the list for search tool development. However, the other side of the usability coin is producing better, more useful results. That beguilingly simple interface still too often reaps a haystack, leaving users wanting for a needle. Luckily, the industry seems to be generating as many approaches to search solutions as there are ways to phrase a query. From full-text searches to metadata, facets, and tag clouds, the techniques continue to evolve and improve, in a quest to deliver more meaning. With the buzz surrounding all things collaborative in Web 2.0, it isn’t a great surprise that social search and other user-affected means of improving search results are soaking up the spotlight. Social search is only just making inroads in the enterprise, but user rankings are familiar tools on consumer sites such as Amazon and Netflix, as are recommendation engines, which leverage these ratings to provide individually optimized product placement. Sounds easy, enough: Get a group of interested users to do the legwork rating things, and then program an algorithm in the background that turns these ratings back on the users as recommendations. But to quote Teen Talk Barbie: “Math is hard.” While getting user groups to rate content, results, or products is a challenge unto itself, mathematically interpreting these results—in ways that produce “you may also like” results that users will, in fact, like—is really tough stuff. Yet this very same collaborative approach may turn out to be the way to improve the technology too. About a year and a half ago, Netflix launched the Netflix Prize, which “seeks to substantially improve the accuracy of predictions about how much someone is going to love a movie based on their movie preferences.” While this may not to be a life-or-death findabilty dilemma, the seemingly modest objective of the competition—a 10% improvement over Netflix’s existing Cinematch solution— and the fairly impressive reward—$1 million for the team that can do it—demonstrate that effective search is not only a complex problem, but a valuable one to solve. As of this writing, the closest any team has come is an 8.82% improvement. This may seem close to the goal, but the rate at which competitors are closing in on 10% slowed to a crawl after the first few months of the competition. In fact, some wonder if the 5 years that Netflix defined as the span of the competition will be enough. The Netflix Prize has done more than demonstrate the challenge and importance of building a better search engine. While a million dollars would be a nice reward for any of the competing teams, it appears that many view the process itself as valuable. One team that held the thirdplace spot early on publicly posted a complete description of its algorithm so that others could build upon it. A member of prominent team “BellKor” (boasting members from AT&T) reportedly said that they already feel rewarded for their efforts by the interaction they’ve had with other teams. With more than 25,000 teams in play for the Netflix Prize, it appears that the search industry will have no shortage of approaches emerging to help better navigate the search problem. One current contest-leader may illustrate this best: Gavin Potter (competing under the name “just a guy in a garage”) does not come at the search business through the usual channels: He is not a programmer, nor a mathematician. He is a self-described “unemployed physiologist,” who is a retired management consultant in the U.K. According to his blog, “I’m fascinated by the use of computers to understand how the mind works and how we can then use this knowledge to help predict human behavior … my attempts at the prize are based around a desire to understand how … to better understand human decision-making.” Mind you, he also writes that he wouldn’t mind winning the million dollars, but something about his approach sends a signal to the search industry as a whole that we must continue to explore a variety of uncharted routes—perhaps as rich in complexity, as difficult to map, and as rewarding as the way ideas connect in the human mind. Michelle Manafy EDITOR-IN-CHIEF 8 ENTERPRISE SEARCH SOURCEBOOK 2008

Table of Contents Feed for the Digital Edition of Enterprise Search Sourcebook 2008

Enterprise Search Sourcebook 2008 Contents Editor’s Note Publisher’s Note Findings and Figures Why Enterprise Search Will Never Be Google-y Searching for Search Usability Your Users Are Talking to You What’s Your Search Story? Search Is Dead—Now What? Delivering on the Promise of Enterprise Search Taming Multiple Search Engines in Your Organization Enterprise Search: Trends for 2008 Enterprise Search Seen From the Inside Open Source Search: Elixir or Poison? Avoiding the Big Mistakes in Search Semantic Search Takes Root in the Enterprise E-Discovery Essentials: The Rules You Need to Know SharePoint Search: An Enterprise Contender? Integrating Security Into Your Enterprise Search Infrastructure Engineering a Better Search Infrastructure Letting End Users Ask the Questions, Stat! The Power of Knowledge Legal Research Using Enterprise Search: A Developer’s View From Treading Water to Full Steam Ahead Pulling Out All the Stops With Midas A Natural Search Solution An Incremental Approach to Improving Enterprise Search The Enterprise Search Sourcebook Showcase Directory Index to Advertisers and Companies Mentioned

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.