# Document-based Question Answering on Human Evaluation 40 document-query pairs

Source: [Document-based Question Answering on Human Evaluation 40 document-query pairs](<https://www.sota2.com/research/sota/document-based-question-answering-on-human-evaluation-40-document-query-pairs>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-25T07:17:08.725Z

Human evaluation of Document-based Question Answering performance across 40 document-query pairs, assessed using metrics like Relevance, Coherence, Completeness, and Answer Quality.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Document-based Question Answering |
| Dataset (including any stated split) | Human Evaluation 40 document-query pairs |
| Metric | Relevance |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Relevance](<https://www.sota2.com/research/sota/document-based-question-answering-on-human-evaluation-40-document-query-pairs.md?metric=Relevance>): Higher is better; unit: SCALAR.

- [Coherence](<https://www.sota2.com/research/sota/document-based-question-answering-on-human-evaluation-40-document-query-pairs.md?metric=Coherence>): Higher is better; unit: SCALAR.

- [Completeness](<https://www.sota2.com/research/sota/document-based-question-answering-on-human-evaluation-40-document-query-pairs.md?metric=Completeness>): Higher is better; unit: SCALAR.

- [Answer Quality](<https://www.sota2.com/research/sota/document-based-question-answering-on-human-evaluation-40-document-query-pairs.md?metric=Answer%20Quality>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Relevance | Result date | Report paper |
| --- | --- | --- | --- |
| QASC | 4.4 | Not recorded | 6a13d77e6f71c5ae7dc9e7e2 |
| Semantic Chunking | 3.8 | Not recorded | 6a13d77e6f71c5ae7dc9e7e2 |
| Agentic Chunking | 3.6 | Not recorded | 6a13d77e6f71c5ae7dc9e7e2 |
| Recursive Splitting | 3.3 | Not recorded | 6a13d77e6f71c5ae7dc9e7e2 |
| Fixed (500) | 3.2 | Not recorded | 6a13d77e6f71c5ae7dc9e7e2 |

### QASC

Relevance: 4.4. Result ID: 6a13f755fbc373592c0ab653.

[Reporting paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Semantic Chunking

Relevance: 3.8. Result ID: 6a13f755fbc373592c0ab651.

[Reporting paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Agentic Chunking

Relevance: 3.6. Result ID: 6a13f755fbc373592c0ab652.

[Reporting paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Recursive Splitting

Relevance: 3.3. Result ID: 6a13f755fbc373592c0ab650.

[Reporting paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Fixed (500)

Relevance: 3.2. Result ID: 6a13f755fbc373592c0ab64f.

[Reporting paper](<https://www.sota2.com/research/paper/6a13d77e6f71c5ae7dc9e7e2.md>)

Recorded evaluation configuration: {"Chunk Size":"500"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.