Skip to main content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Text REtrieval Conference (TREC) 25th Anniversary Symposium

Thanks, Ellen. Welcome to everyone here in the auditorium and watching the webcast.

I’m happy to be here to celebrate 25 years of TREC and the great progress this conference has generated in the text retrieval field.

When TREC started in 1992, the World Wide Web connected all of 26 servers. Back then, no one could have imagined how quickly the web would expand to become the world’s ubiquitous information resource.

Nor could anyone have predicted in 1992 how central the success of text retrieval and search engines would be to modern life today.

TREC first started with 25 organizations testing their algorithms with the newly created NIST text collection.

Today, TREC has helped more than 300 organizations on six continents benchmark their newest text retrieval and searching tools.

NIST is grateful to Ellen Vorhees for her steady, capable leadership of the TREC effort for the last 20 years.

We’re also grateful to the DARPA for their support to NIST’s budding text retrieval group back in 1990. They asked us to create a test collection of about 1 million full-text documents, which was 100 times larger than any other publicly accessible text database at the time.

If you have been coming to this conference for many years, you know that the benchmarking exercises conducted through TREC have become the defacto standard by which text retrieval algorithms were measured.

A 2010 economic study conducted by RTI International found that for every $1 invested by NIST and its partners in TREC, $3 to $5 dollars in benefits were created for text retrieval researchers.

This impact has come in three main areas:

  • First, TREC has continually pushed the state-of-the-art for document retrieval. Since the start of the conference, TREC participants have doubled the useful documents returned for a given search.
  • Second, it has been a reliable way to spread new approaches. Most commercial search engines now use technology first debuted at TREC conferences.
  • And third, TREC has been instrumental in expanding research into new areas that can benefit from better text retrieval methods.

TREC expands research with discussion tracks that change each year to keep up with current trends. Recent additions have included tracks for

  • posing search requests in one language and retrieving documents in another,
  • for finding video resources based on content,
  • and for conducting relevant searches that answer questions rather than using key words.

And as if that’s not enough, TREC even helped win a million dollars. Feedback IBM received through TREC’s question-answering track helped fine tune its “Jeopardy” playing computer. In 2011, Watson used its new skills to defeat two human champions of the game.

The impact of TREC is likely to grow as we tackle even more difficult challenges such as Artificial Intelligence, or AI.

In October, the White House Office of Science and Technology Policy released a report concluding that artificial intelligence “can be a major driver of economic growth and social progress.” Possible applications for AI include everything from improved medical diagnoses to better air traffic control to enhanced real-time detection of cybersecurity threats.

TREC can play a role by convening cross-disciplinary conversations in areas where search functions and AI intersect.

A key opportunity in AI is to improve the ability of machines to process natural language rather than highly structured data or information. When a robot, for example, can process language more like a person, it will allow AI to tackle harder problems and advance the related field of machine learning.

Natural language, of course, is what text retrieval and search algorithms are all about. It’s exciting that TREC is playing a role in this new world.

We have a full program for you today. And you’ll hear from other speakers who have directly benefited from TREC. I hope you will find their words inspiring for your own work.

Searching is a human need. It’s as old as our species, which makes the progress made in just the last 25 years even more remarkable.

Many of the people here made that possible, so please accept our gratitude for your dedication to this community.

I look forward to hearing from Ellen about your latest accomplishments. Thank you.

Created November 15, 2016