This series of meetings originated in Albuquerque, New Mexico in 1995. That inaugural meeting (part of an ASIDIC series) was transplanted to Bath in England (1996 and 1997) and then to Boston, Massachusetts (1998, 1999 and 2000). The Search Engines Meetings bring together commercial search engine developers, academics and corporate professionals to learn from each other. Infonortics, sponsor of meetings post-1995 with Ev Brenner, will continue the same success in Boston in 2001. Early booking is recommended; these meetings get crowded.
Book now for the 2001 Search Engines Meeting
(secure server)
See also details of
how to exhibit at this meeting
Last updated 6 April 2001
Day One Keynote - 09.05-09.50
David Seuss (Northern Light,
Massachusetts)
Search Strategies for Large Enterprises
The Web is bringing profound changes in the electronic information
available to individuals. Everyone uses Web search engines in
their business, personal and investing lives. As a result, everyone
has become his or her own research librarian and expects to seek
and use electronic information on each and every task. Returns
to individuals and to enterprises from having information resources
and effective search capability are soaring, and search has become
a central focus of many new business-critical applications.
Unfortunately, the evolving new class of search applications requires
technology, library skills, content relationships and operating
infrastructure that few companies have in their IT departments.
_________________________________________________________
Panel: Search Engine
Developments and Trends since April 2000 - 09.50-10.50
Greg Notess (Montana State University-Bozeman) - http:/searchengineshowdown.com/
Chris Sherman (About.com, California) - http://websearch.about.com/
Avi Rappoport (Searchtools.com) - www.searchtools.com/
Two noted commentators on the search engine scene present their analyses of what we have achieved over the past twelve months, and where we are heading next
Networking break 10.50-11.15
Lou Rosenfeld (Argus Center
for Information Architecture, Michigan) 11.15-11.50
We've Got it Backwards: an information architect's viewpoint
for broadening searching and narrowing domains
The search engine community has made gargantuan efforts to
provide improved access to broad, complicated collections of information
that are used by highly heterogeneous audiences. These are difficult
challenges and, not surprisingly, lead to less than ideal information
retrieval performance.
The solution may not lie in using better search algorithms, but
in re-examining the problem itself. First, users typically do
not care about search engines and how they work; they just care
about getting the information they want. Second, the more broadly
we define search spaces, the more difficult it is for conventional
search solutions to succeed. So we should apply the holistic finding
approach to specialized problems where content is more homogeneous,
less voluminous, and where users and their information needs are
better understood.
Bob Travis (Altavista,
California) 11.50-12.25
The Need Behind the Query: Web Search vs Classic Information
Retrieval
Classic IR is predicated on users searching for needed information.
But the intent behind a web search is often not informational
-- it might be navigational (show me the url of the site I want
to reach) or transactional (show me sites where I can perform
a certain transaction, eg shop, download a file, or find a map).
This presentation dissects some of the differences between classic
IR and web search and shows how search engines evolved to deal
with web-specific needs.
Chris Cardinal (Hummingbird,
Canada) 12.25-1.00
The TREC and commercial search companies
Commercial participation in TREC has been almost non-existent.
This year Hummingbird participated for the first time, submitting
runs in the web and query tracks using the Fulcrum search engine.
Hummingbird will talk about how it approached TREC and the value
derived from this experience.
Lunch: 1.00-2.15
Mark Hansen and Elizabeth Shriver
(Bell Labs, Lucent Technologies,
New Jersey) 2.15-2.50
The Information Hidden When Web Searching
We propose techniques for both personalizing
and improving the effectiveness of searching by focusing on the
presentation and ranking of search results. We display groups
of URLs, where the URLs in a group are "related". We
determine the relationship by applying several types of model-based
clustering to the key words of the query as well as the web pages
visited during the search session and the timing of each session.
We trained a system with the search sessions from a number of
months of Bell Labs proxy logs, and verified the usefulness of
our clustering scheme with a few months of new search session
data. In most cases, we report savings both in terms of the number
of URLs visited as well as the time required for a search.
Panel: Content Architecture
(Part I) 2.50-4.00
Chairman: Susan Feldman, IDC, Massachusetts
Is search terrible, as the headlines proclaimed last fall, or
is it the materials that search engines are crunching? What can
we do to improve the situation? These sessions explore technologies
that can improve search. We look at techniques that help structure
unstructured text such as categorization, metatags, XML, entity
extractions, and authority files. We examine some of the interaction
techniques that will guide searchers to the information they are
seeking. What is still a dream, and what is now available on the
market? Each speaker will explore one facet of how to make search
better.
Sue Feldman (IDC, Massachusetts) -- Introduction: What is content
architecture?
Mark Bennett (Searchbutton, California) -- XML
Richard Boulderstone (Looksmart, California) -- metadata
Eytan Ruppin (Zapper Technologies, Israel) -- putting search and
categorization in context
Networking Break: 4.00-4.25
Panel: Content Architecture (Part II) 4.25-5.45
Daniel Lulich, (Rulespace, Oregon) -- automatic classification
and its visualization
Tom Wilde (FAST, MA) -- scaling and content aggregation platforms
Horst Koerner (Seruba, Germany) -- cross language ontologies and
topic maps - knowledge engineering
Ashok Chandra, (Verity, California) -- transforming the web into
social networks
07.30-08.25 -- Breakfast
Day Two Keynote - 8.30-9.10
Eric Brewer (Inktomi Corporation,
California)
The Next Evolution: A New Infrastructure for Connecting Content
with Users
Dr Eric Brewer, co-founder and chief scientist of Inktomi, will discuss how an emerging infrastructure for content providers connecting to search engines will enable users to access higher quality results when searching and browsing the Web. He will also examine the new business and technical arrangements that are being developed to enable content providers to make their best and most current content readily available to consumers.
________________________________________________________________________________________________
Bernard Normier (LexiQuest,
New York) 9.10-9.45
From String Indexing to Semantic Indexing. The example
of Lexisez, an NLP-based question / answering system
Question Answering systems are a new application in the general framework of information retrieval. Several strategies are possible, from question/answers databases, to answer extraction from plain text. On the search engine part of the system, the indexing is not controlled, and the query is expanded to a Boolean query using NLP analysis of the query and semantic expansion. On the question/answer database, the index is the result of a linguistic analysis. Two levels of indexing are used: classical part of speech level, and a more sophisticated semantic level. At the semantic level, the indexes are no more words, but "concepts", that is words are disambiguated and a given meaning of the word is selected. When not possible, the sentence is over-indexed by several meanings. At this level too, the index is not a flat list of descriptors, but a structured representation of the basic form. A sophisticated matching algorithm is then used, which improves in large proportions both precision and recall.
David A Evans (Clairvoyance,
Pennsylvania) 9.45-10.20
The 'Affect' Dimensions of Search
Search technology and functionality have traditionally focused on the problem of matching queries to documents, principally by comparing the content of the query to the content of documents. Many internet-based applications do not involve literal queries but do require a 'match' between the content of a text and some other information object. An example of such an application is the matching of a banner ad or offer to a user who is browsing through a web page with certain content; or the suggestion to a user of a book, or video, or CD based on the content or description of another book, etc, that the user may be reading. Especially in such contexts, the superficial similarity of texts may be a poor guide to appropriate "matches". Another dimension -- affect (or emotion) -- provides a much better measure of the features that matter in making good suggestions. This talk will describe one approach to the identification, measurement, and use of the multiple dimensions of affect in texts to make internet search and related applications more effective.
Networking break: 10.20-10.40
Stephen E. Arnold (AIT,
Kentucky) 10.40-11.15
Vertical Search Engines
Automated systems for indexing the Web are useful. The surge in topic or market specific Web sites has created a need for more restricted collections of Web pointers and indexes to high value content closely related to the focus of a Web site. The vertical collection functionality of four search engines will be described and illustrated. The engines included in the presentation are:
· Zbase, a new automated service that builds a Yahoo!
style directory
· EoExchange, an engine that has been tuned to create vertical
content collections
· AnswerSleuth, a combination of human editorial and automated
content indexing
· Mohomine, an engine that creates vertical collections
from Web sources as well as selected content created by employees
of an organization.
The "best fit" for each of these vertical search engine services will be identified. General pricing parameters for each service indicate the direction that vertical search services are moving in the internet / intranet markets.
John Snyder, Webtop.com,
Cambridge, England 11.15-11.50
Beyond the Keyboard: Speech Enabled Search
Speech recognition is increasingly entering our consumer lives
as a call direction technology and dictation tool. How can the
latest technology be harnessed for speaking real natural language
searches? Dividing the Web into zones of content assists the delivery
of untrained "speaking search engines" to a new generation
of mobile web users.
Matthew Koll (Virginia)
11.50-12.25
Massively Distributed Search: Still a Good New Idea
The invisible web continues to elude the spiders. Coverage and precision of web searches are not as good as we would like. Timeliness matters. Cost of service matters. Business imperatives impact the supply side. P2P gives a push -- but is it just another push? When will massively distributed search happen?
Lunch: 12.25-1.55
Craig Silverstein, (Google, California) 2.00-2.35
Beyond Search Engines: The Future
History of Information Retrieval
The pace of advanced search technology is accelerating. What does 'search' mean in this brave new world? This presentation discusses Google's view of one way in which search could develop. With search sites becoming the destination of first result for Internet users, there will be more emphasis on quality, from 'search' to 'find'. User interfaces must be able to simplify query inputs, synthesize results, and return only the most relevant information. The presentation will also provide a perspective on what search features will likely have the biggest impact on the search industry, and why.
Gregory T Grefenstette
(Xerox Research Centre Europe, France) 2.35-3.10
Multilinguality in the Web
The Web is and will remain multilingual. We will show how language is automatically identifiable by search engines, how one can estimate the volume of a language accessible through a search engine, and what language-specific resources are needed to do simple, but good indexing.
Networking break 3.10-3.30
Panel: The Secularization of Search - 3.30-4.45
Chairman: David A. Evans (Clairvoyance, Pennsylvania)
William R. Hersh (Oregon Health Sciences University (Oregon)
Joshua Arai and Keiichi Kitagawa (Justsystem Corporation, Japan)
Before 1992, there were probably fewer than one million people worldwide -- mostly professionals in library or information science, or scientists and physicians --who had ever conducted an automated text search. By 1995, there were probably 100 million -- "ordinary" people who had searched the internet via their browsers and a search engine. With search now unambiguously in the hands of non-professionals and with the use of search now well integrated into regular business and personal use of the internet, we might well ask whether a "received view" of search is emerging. This panel will explore this question by examining several cases from around the world in which search as a discipline is being taught in schools. In each case, we will attempt to answer a series of questions: Why is search perceived to be important? What aspects of search are in focus? How is the methodology taught? What is the expected outcome and how is it measured?
The meeting fee is US $695 and is payable in advance. The fee includes the two day meeting, two breakfasts, two lunches, the meeting cocktail, and all breaks. Please note meeting fees are payable in respect of all reservations for this event unless a cancellation has been received under the terms below. For those working at an academic institution, there is a limited number of places at a special academic rate of US $495. The fee for the half-day pre-conference workshops on April 8 is $290 per workshop. The workshops are held in parallel, so you will only be able to attend one per person.
Organisations will receive a discount of 15% on all bookings if more than two bookings from the same organisation are received. Bookings for multiple attendees must be received at the same time and be invoiced on the same invoice. This discount is not in addition to the academic or early booking rates (standard rate only applies).
Notice of cancellations must be received at least ten working days prior to the event (ie, by March 26); a cancellation, refund and handling fee of $80 will be charged. Cancellations notified after this date cannot be accepted for refund, and the full meeting fee will stand. Substitutions may be made at any time.
Hotel accommodation is not included in the meeting price. The meeting hotel, the prestigious Fairmont Copley Plaza, has a limited block of rooms available for attendees at this meeting. The special meeting rate is US $219.00 (excluding tax). Those booking should mention "Search Engine Meeting" to be given this special rate. Once this room block is fully occupied, normal hotel rates may apply.
Registrations may be made at the meeting registration desk at the Fairmont Copley Plaza on Sunday April 8 from 4 pm until 6.30 pm, or on Monday April 9 from 8 am. Workshop registrations are Sunday April 8 from midday.
The meeting sessions begin on Monday April 9 at 8.45 am. The meeting ends Tuesday April 10 at approximately 5.15 pm.
The meeting mixer cocktail will take place the evening of Monday April 9 at approximately 6 pm.
The meeting is at the Fairmont Copley Plaza Hotel, 138 St James
Avenue, Boston, MA 02116, USA.
Telephone is +1 617 267 5300. Fax is +1 617 375 9648. Toll-free
reservation phone within USA is: 1-800-527-4727.