The cost of litigation in the US involving electronically stored information (so-called “E-discovery”) is burgeoning: according to one Forrester study issued in 2006, corporate America is expected to increase its spending on E-discovery from what was $1.4 billion dollars in 2006 to $ 4.8 billion in 2011. Under the new US Federal Rules of Civil Procedure effective December 1, 2006, both private and public sector litigants increasingly confront a world of preservation orders, injunctions, subpoenas, and other demands for access to exponentially increasing volumes of records and information stored electronically. One set of emerging best practices highlighted by a recent best practices commentary published by The Sedona Conference® involves more serious attention being paid by lawyers and judges to information retrieval issues, and the deficiencies in the way in which documents and ESI routinely are searched for, including by way of “keywords.” This presentation provides a strategic approach to search problems lawyers and their clients face in e-discovery, as well as practical pointers drawn from real cases; it also includes the latest findings from the TREC Legal Track, an international text retrieval project run out of the US National Institute of Standards and Technology.

Presentation - PDF Document (555 KB)

Article - PDF Document (234 KB)

This presentation discusses trends and benefits of the convergence of search and visualization tools, including brief historical information to provide context. It looks at the latest tools as well as reviews the early adoption and specific use cases in a number of verticals. Finally it discusses how powering search results with visualization will impact the market moving forward, including the impact on today's most popular business models.

Marcelline Saunders - PDF Document (2 MB)

Following from research, observations, and interviews we determined that search and sense-making involved many different component tasks and many different workflows through these tasks. Advanced technologies, such as entity extraction, classification, etc, address part of the problem, but significant end-user performance improvement in search and sense-making tasks can be also be achieved through innovating the end-user interface to enable better workflows across these different technologies and tasks with a single unified interface. We have created a new interface using visualization techniques called nSpace to implement these ideas including a component called TRIST for interaction with large amounts of results data to help users find the relevant, novel and unexpected; and a component called Sandbox for collecting, organizing and reasoning with pieces of information for sense-making. nSpace uses novel techniques such as linked dimensions for characterization and use of gestures for fluid workflows. The presentation discusses some of the research findings, shows some examples of the nSpace interface, and discusses some of the results and feedback.

Richard Brath - PDF Document (3 MB)

Enterprise Search Software as it is known today, whether from Autonomy, Endeca, FAST or others, cannot provide access to all the information of value at any reasonably sized organization with a single search. Organizational information-content exists in numerous silos accessible through a myriad of individual, incompatible indices-engines. Technical, cost and bureaucratic reasons prevent unifying all these various enterprise silos under one index.

This presentation discusses how state-of-the art Federated Search software provides actual enterprise (-wide) single point of search-access to most, if not all, of the information repositories of value to an enterprise, including those beyond the firewall.

Abe Lederman - PDF Document (730 KB)

Recent trends in search have centered largely on specialized search functions, such as image and video search, signaling the importance of the ability to index specific types of information and different types of content. However, search’s potential does not end there. It will continue to grow, providing users with the ability to combine different forms of content in more effective ways. By connecting sources that were until now considered separate, hybrid/vertical search eclipses traditional text-oriented and directory-based search methods, and is poised to become the search industry’s next big trend.

In this presentation, Bourdoncle offers a technical perspective on the latest technologies to facilitate hybrid/vertical search, which ultimately fosters a simpler, more natural search experience for the user. Bourdoncle also provides insight into his first-hand experience in helping to design Exalead’s hybrid/vertical search solution and will discuss the various opportunities for employing hybrid/vertical search. In addition, the presentation addresses the benefits and challenges of the technologies employed, including entity extraction, real-time indexing, taxonomies and navigation.

Spencer Shearer - PDF Document (1 MB)

Is Enterprise Search the dull cousin of web-wide search? Is it destined to play follow-the-leader in the innovation game, gradually adapting technologies to the enterprise that were originally created for full web searches by Google, Yahoo, and MSN?

This presentation suggests that Enterprise Search is not just one application of full web search technology, but rather a fundamentally different problem. As such, the technologies of Enterprise Search and full web search are on unavoidably divergent paths; differences in content production, quality, and knowledge of domain will inevitably lead to fundamentally different solutions for the two problems.

Edwin Cooper - PDF Document (1 MB)

Enterprise IT managers are increasingly looking for the quickest and most effective ways to bring the best of Web 2.0 and social networking into their organizations. This is a difficult task and many IT managers do not know where to where to begin and which technologies offer the fastest return on investment in this brave new world. One technology that is showing great promise in terms of bringing together the best of Web 2.0 for the enterprise is relational navigation, because it can bring together the best of the Web, such as tagging, sharing and annotating results to extend the huge investments that have already been made in enterprise content management (ECM), document and database management, and enterprise search.

Fundamentally, relational navigation is about providing a more effective way to find and manage the myriad of content enterprises import, store and export. It improves ECM by leveraging semantic technology and the principles of the social Web to aggregate, organize, manage and navigate an information centric architecture in ways that were never possible before. By focusing on the relationships between people, places and things, relational navigation maintains context and allows participation in the discovery process. In this presentation attendees will hear:

• The benefits of relational navigation

• How Web 2.0 is changing the enterprise

• The future of Web 2.0

Brad Allen - PDF Document (2 MB)

Open Search, an new XML standard, has simplified the process of standardizing search results from multiple sources. Open Pipeline is a new standard for the indexing side of the equation. It is a simple, common architecture for connectors to data sources, file filters, text analyzers, and modules to distribute documents across a network. Partly an API and partly a feed protocol based on Atom/RSS, it provides an open, non-proprietary way to fetch, parse, analyze, and route documents.

Chris Cleveland - PDF Document (4 MB)

In today’s fast moving marketplace, content applications need to quickly adapt to new content sources and new market demands. MarkLogic Server’s XML classifier is a new tool to let content owners:

• Classify XML at any level of granularity: assign class membership for the whole document, an individual element, or anywhere in between

• Classify synthetic documents: use the title, abstract, and first paragraph of each section, ignoring the footnotes

• Classify based on semantics or syntax: leverage indexes for text, structure, or a mixture of both

• Incorporate classification into the rendering logic: allow classification output to dynamically affect the rendering of content

This presentation describes how MarkLogic exposes its XML classifier through an XQuery interface, and shows a demonstration of the XML classifier at work within a content application built using MarkLogic.

Kelly Stirman - PDF Document (240 KB)

Two expert industry commentators reflect on what was said during the two days of the 2008 Search Engine Meeting and, with the help of the audience, draw some lessons and conclusions.