# Telecom Fraud Analysis on TeleAntiFraud-Bench

Source: [Telecom Fraud Analysis on TeleAntiFraud-Bench](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench>)

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

Data updated: 2026-02-26T06:08:47.928Z

Evaluation of Telecom Fraud Analysis performance using the TeleAntiFraud-Bench dataset, reporting weighted F1 scores across different fraud categories and qualitative assessment metrics.

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 | Telecom Fraud Analysis |
| Dataset (including any stated split) | TeleAntiFraud-Bench |
| Metric | Weighted F1 (Sce.) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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

- [Weighted F1 (Sce.)](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Weighted%20F1%20(Sce.)>): Higher is better; unit: PERCENT.

- [Weighted F1 (Fra.)](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Weighted%20F1%20(Fra.)>): Higher is better; unit: PERCENT.

- [Weighted F1 (FT.)](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Weighted%20F1%20(FT.)>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

- [Logical Rigor](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Logical%20Rigor>): Higher is better; unit: SCALAR.

- [Practical Value](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Practical%20Value>): Higher is better; unit: SCALAR.

- [Expression Quality](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Expression%20Quality>): Higher is better; unit: SCALAR.

- [Total Qualitative Score](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Total%20Qualitative%20Score>): Higher is better; unit: SCALAR.

- [Final Score](<https://www.sota2.com/research/sota/telecom-fraud-analysis-on-teleantifraud-bench.md?metric=Final%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Weighted F1 (Sce.) | Result date | Report paper |
| --- | --- | --- | --- |
| Deepseek V3 | 88.53 | Not recorded | 695c96e353fdd5c361743413 |
| SAFE-LS | 84.64 | Not recorded | 695c96e353fdd5c361743413 |
| Deepseek R1 | 83.6 | Not recorded | 695c96e353fdd5c361743413 |
| SAFE-RS | 81.6 | Not recorded | 695c96e353fdd5c361743413 |
| SAFE-RL | 81.57 | Not recorded | 695c96e353fdd5c361743413 |
| AntiFraud-Qwen2Audio | 81.31 | Not recorded | 695c96e353fdd5c361743413 |
| Gemini-2-Flash | 80.51 | Not recorded | 695c96e353fdd5c361743413 |
| GPT4-o | 80.25 | Not recorded | 695c96e353fdd5c361743413 |
| InternLM2.5-20B | 78.34 | Not recorded | 695c96e353fdd5c361743413 |
| Qwen2.5-72B | 78.31 | Not recorded | 695c96e353fdd5c361743413 |
| Step-Audio-Chat | 76.35 | Not recorded | 695c96e353fdd5c361743413 |
| GLM4-9B-Chat | 75.1 | Not recorded | 695c96e353fdd5c361743413 |
| Doubao 1.5 Pro | 71.14 | Not recorded | 695c96e353fdd5c361743413 |
| Qwen2-Audio-7B-Instruct | 70.22 | Not recorded | 695c96e353fdd5c361743413 |
| GLM4-9B-Voice | 0 | Not recorded | 695c96e353fdd5c361743413 |

### Deepseek V3

Weighted F1 (Sce.): 88.53. Result ID: 699a0a7f5698665ec48d7d93.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77001e209131cea7066a.md>)

Recorded evaluation configuration: {"Type":"ASR+LLM"}. This record does not assert independent reproduction.

### SAFE-LS

Weighted F1 (Sce.): 84.64. Result ID: 699a0a7f5698665ec48d7d9d.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"LALM+Ours"}. This record does not assert independent reproduction.

### Deepseek R1

Weighted F1 (Sce.): 83.6. Result ID: 699a0a7f5698665ec48d7d94.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76a91e209131cea6e3eb.md>)

Recorded evaluation configuration: {"Type":"ASR+LRM"}. This record does not assert independent reproduction.

### SAFE-RS

Weighted F1 (Sce.): 81.6. Result ID: 699a0a7f5698665ec48d7d9c.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"LALM+Ours"}. This record does not assert independent reproduction.

### SAFE-RL

Weighted F1 (Sce.): 81.57. Result ID: 699a0a7f5698665ec48d7d9b.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"LALM+Ours"}. This record does not assert independent reproduction.

### AntiFraud-Qwen2Audio

Weighted F1 (Sce.): 81.31. Result ID: 699a0a7f5698665ec48d7d9a.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7aa41e209131cea84503.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

### Gemini-2-Flash

Weighted F1 (Sce.): 80.51. Result ID: 699a0a7f5698665ec48d7d96.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

### GPT4-o

Weighted F1 (Sce.): 80.25. Result ID: 699a0a7f5698665ec48d7d97.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

### InternLM2.5-20B

Weighted F1 (Sce.): 78.34. Result ID: 699a0a7f5698665ec48d7d90.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"ASR+LLM"}. This record does not assert independent reproduction.

### Qwen2.5-72B

Weighted F1 (Sce.): 78.31. Result ID: 699a0a7f5698665ec48d7d91.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"ASR+LLM"}. This record does not assert independent reproduction.

### Step-Audio-Chat

Weighted F1 (Sce.): 76.35. Result ID: 699a0a7f5698665ec48d7d98.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77011e209131cea706b5.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

### GLM4-9B-Chat

Weighted F1 (Sce.): 75.1. Result ID: 699a0a7f5698665ec48d7d8f.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"ASR+LLM"}. This record does not assert independent reproduction.

### Doubao 1.5 Pro

Weighted F1 (Sce.): 71.14. Result ID: 699a0a7f5698665ec48d7d92.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

Recorded evaluation configuration: {"Type":"ASR+LLM"}. This record does not assert independent reproduction.

### Qwen2-Audio-7B-Instruct

Weighted F1 (Sce.): 70.22. Result ID: 699a0a7f5698665ec48d7d99.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74281e209131cea6017e.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

### GLM4-9B-Voice

Weighted F1 (Sce.): 0. Result ID: 699a0a7f5698665ec48d7d95.

[Reporting paper](<https://www.sota2.com/research/paper/695c96e353fdd5c361743413.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75f91e209131cea6abba.md>)

Recorded evaluation configuration: {"Type":"LALM"}. This record does not assert independent reproduction.

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