LetTokenBurnWeb Data Commones - Product Attribute Value Extraction
The datasets contains 1,420 human annotated product offers, systematically selected from the Web Data Commons Product Matching Corpus, featuring 24,582 annotated attribute-value pairs, making it a valuable resource for both product attribute-value extraction and product matching tasks.
$ npm i -g lettokenburn-cli$ ltb bench info wdc-pave
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| GPT-4_10_example_values_&_10_demonstrations | 90.5 | N/A | Independent | medium | Mar 4, 2024 | arxiv.org |
| GPT-3.5_10_example_values_&_10_demonstrations | 88.0 | N/A | Independent | medium | Mar 4, 2024 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
The datasets contains 1,420 human annotated product offers, systematically selected from the Web Data Commons Product Matching Corpus, featuring 24,582 annotated attribute-value pairs, making it a valuable resource for both product attribute-value extraction and product matching tasks. The normalized gold standard contains the standardized attribute value pairs as described below.