Text Retrieval Conference

src: s3.amazonaws.com

The Text REtrieval Conference ( Trec ) is a series of ongoing workshops focusing on a list of different areas of information retrieval (IR), or track. It is co-sponsored by the National Institute of Standards and Technology (NIST) and the Intelligence Advanced Research Project Activities (part of the office of the Director of National Intelligence), and started in 1992 as part of the TIPSTER Text program. The goal is to support and encourage research in the information retrieval community by providing the necessary infrastructure for large evaluation of text-capture methodologies and to increase the speed of transfer of lab-to-product technology.

Each track has challenges where NIST provides participating groups with data sets and test issues. Depending on the track, the test issue may be a question, topic, or feature target that can be extracted. A uniform assessment is conducted so that the system can be evaluated fairly. After the results evaluation, a workshop provides a place for the participants to collect thoughts and ideas together and present current and future research work.

Video Text Retrieval Conference

Tracks

Current tracks

In 1997, the Japanese partner of TREC was launched (the first workshop in 1999), called NTCIR (NII Test Collection for IR Systems), and in 2000, the CLEF, a European partner, specifically illustrated by the study of cross-language information retrieval is launched.

Maps Text Retrieval Conference

Conference contributions for search effectiveness

NIST claims that in the first six years of the workshop, the effectiveness of the retrieval system is approximately twice that. The conference was also the first to conduct a large-scale evaluation of non-English documents, speeches, videos, and cross-language retrieval. In addition, the challenge has inspired a large number of publications. The first technology developed at TREC is now included in many of the world's commercial search engines. An independent report by RTII found that "about a third of the increase in web search engines from 1999 to 2009 was due to TREC, possibly storing up to 3 billion hours of time using web search engines.... In addition, reports indicate that for every $ 1 which NIST and its partners invested in TREC, at least $ 3.35 to $ 5.07 in benefits gained for US information searchers both in the private sector and academia. "

While one study shows that the state of the art for ad hoc search has not grown substantially in the last decade, it refers only to searching topically relevant documents in small news and web collections of several gigabytes. There have been advances in other types of ad hoc searches in the last decade. For example, a test collection was created for a known web search item that found an increase from the use of anchor text, title weighting and url length, which is not a useful technique in older ad hoc test collections. In 2009, a collection of billions of new web pages was introduced, and spam filtering was found to be a useful technique for ad hoc web search, unlike in previous test collections.

The collection of tests developed at TREC is useful not only to (potentially) help researchers advance the state of the art, but also to enable new (commercial) product developers to evaluate their effectiveness on standardized tests. In the past decade, TREC has created new tests for enterprise e-mail search, genomic search, spam filtering, e-Discovery, and several other retrieval domains.

The TREC system often provides the basis for further research. Examples include:

• Hal Varian, Chief Economist at Google, says Better data yields better science. The history of information retrieval describes this principle well, "and describes TREC's contribution.
• The TREC Legal Track has affected the e-Discovery community both in research and in the evaluation of commercial vendors.
• The IBM research team built IBM Watson (aka DeepQA), which beat the world's best Jeopardy! players, using data and systems from TREC's QA Track as a basic performance measurement.

src: pbs.twimg.com

Participation

The conference is made up of a diverse group of international researchers and developers. In 2003, there were 93 groups from academia and industry from 22 participating countries.

src: i0.wp.com

References

src: cdn2.engineerica.com

External links

• TREC website at NIST
• TIPSTER
• TREC book (in Amazon)

Source of the article : Wikipedia