MétaCan
Menu
Back to cohort
Record W2294145134

Overview of the TREC 2009 Web Track

2009· article· en· W2294145134 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenueText REtrieval Conference · 2009
Typearticle
Languageen
FieldComputer Science
TopicInformation Retrieval and Search Behavior
Canadian institutionsUniversity of Waterloo
Fundersnot available
KeywordsComputer scienceInformation retrievalRelevance (law)Task (project management)World Wide WebClefSession (web analytics)Web pageSet (abstract data type)Search engine indexingQuery expansion
DOInot available

Abstract

fetched live from OpenAlex

The TREC Web Track explores and evaluates Web retrieval technology over large collections of Web data. In its current incarnation, the Web Track has been active since TREC 2009, where it included both a traditional adhoc retrieval task and a new diversity task [4]. The goal of this diversity task is to return a ranked list of pages that together provide complete coverage for a query, while avoiding excessive redundancy in the result list. For TREC 2010 the track introduced a new Web spam task and Web-style, six-level relevance assessment for the adhoc task [5]. For TREC 2011, as recommended by participants at the track planning session held during TREC 2010, we dropped the spam task but continued the other tasks essentially unchanged. As we did for TREC 2009 and TREC 2010, we based our TREC 2011 experiments on the billion-page ClueWeb09 collection created by the Language Technologies Institute at Carnegie Mellon University. The tasks use a common topic set, differing only in their evaluation methodology. Topics are created from the logs of a commercial search engine, with the aid of tools developed at Microsoft Research [9]. Given a target query, these tools extract and analyze groups of related queries, using co-clicks and other information, to identify clusters of queries that highlight different aspects and interpretations of the target query. These clusters are employed by NIST for topic development. Each resulting topic is structured as a representative set of subtopics, each related to a different user need. The selection of subtopics attempts to reflect a mix of genuine user requirements for the topic. For the adhoc task documents are judged with respect to the topic as a whole. Relevance levels are similar in structure to the levels used in commercial Web search, including a spam/junk level. Moreover, the top two levels of the assessment structure are closely related to the homepage finding and topic distillation tasks appearing in older Web Tracks. For the diversity task, documents are judged with respect to the subtopics, as well as with respect to the topic as a whole. For TREC 2011, the topic selection process was modified slightly from previous years. For TREC 2009 and 2010, topics were chosen to be of medium-to-high frequency. TREC 2011 attempts to work with more obscure topics, which may still be underspecified (i.e., faceted) but may be less ambiguous. Search engines have difficulty with queries of this type, since they can rely less on click/anchor information, and popularity signals like PageRank. With these new tough topics we hope to work in an area of Web retrieval that has received relatively little attention. Given the smaller number of pages that may be relevant for these tough topics, we may potentially be able to create a more reusable collection, with sufficiently complete judgments.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.000
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Theoretical or conceptual · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.556
Threshold uncertainty score0.404

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0000.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.001
Science and technology studies0.0000.000
Scholarly communication0.0000.001
Open science0.0020.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.071
GPT teacher head0.309
Teacher spread0.238 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it