Codesota · Models · SumHiSSumHiS Authors3 results · 1 benchmarks
Model card

SumHiS.

SumHiS Authorsopen-sourceExtractive summarization exploiting hidden document structure

SumHiS: Extractive Summarization Exploiting Hidden Structure. arXiv:2406.08215 (Jun 2024). Uses latent sentence-graph structure for extractive selection. Achieves SOTA ROUGE-2 (32.52) among extractive models on CNN/DM by exploiting paragraph-level hidden structure with semantic filtering.

§ 01 · Benchmarks

Every benchmark SumHiS has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01cnn-/-daily-mailComputer Vision · Optical Character Recognitionrouge-232.5%#1/332024-06-12source ↗
02cnn-/-daily-mailComputer Vision · Optical Character Recognitionrouge-l42.4%#5/332024-06-12source ↗
03cnn-/-daily-mailComputer Vision · Optical Character Recognitionrouge-143.5%#16/332024-06-12source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area

Where SumHiS actually performs.

Computer Vision
1
benchmark
avg rank #7.3
§ 03 · Papers

1 paper with results for SumHiS.

  1. 2024-06-12· Natural Language Processing· 3 results

    SumHiS: Extractive Summarization Exploiting Hidden Structure

§ 05 · Sources & freshness

Where these numbers come from.

arxiv
3
results
3 of 3 rows marked verified.