University of Waterloo at TREC 2014 Contextual Suggestion: Experiments with suggestion clustering
Bibliographic record
Abstract
Abstract : In this work we present our group's first attempt at developing a system to solve the problem presented in the contextual suggestion task. As part of TREC 2014 the contextual suggestion track is running for the third time. The goal of this task is to tailor point-of-interest suggestions to users according to this preferences. Here we present how we gathered candidate points-of-interest, grouped them according to similarity using clustering, and picked points-of-interest that each user would nd especially appealing. The organizers of this track distributed users' personal pro les in three les: examples2014.csv, pro les2014-70.csv and pro les2014-100.csv. A list of 100 example points-of-interest, which each consist of an ID, a title, a description and a URL were included in examples2014.csv. 299 users indicated their preferences by giving a rating on a 5-point score (0, 1, 2, 3, 4) to the description and website of each example point-of-interest. 116 users, indicated preferences to all the 100 example points of interests, these pro les are distributed in pro les2014-100.csv. The other 183 users, only indicated 70% of all the example points of interest, these pro les are distributed in pro les2014- 70.csv. There are 50 contexts which each represent a city in the United States which are listed in contexts2014.txt.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.009 | 0.035 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.002 | 0.004 |
| Science and technology studies | 0.004 | 0.001 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.013 | 0.005 |
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.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".