Update README.md
Browse files
README.md
CHANGED
|
@@ -56,6 +56,61 @@ If you find LLM-based topic generation has hallucination or instability, and coh
|
|
| 56 |
}
|
| 57 |
```
|
| 58 |
|
| 59 |
-
(Other citations omitted for brevity, but should remain in the final PR)
|
| 60 |
|
| 61 |
-
If you
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
}
|
| 57 |
```
|
| 58 |
|
|
|
|
| 59 |
|
| 60 |
+
If you use the human annotations or preprocessing:
|
| 61 |
+
```
|
| 62 |
+
@inproceedings{li-etal-2024-improving,
|
| 63 |
+
title = "Improving the {TENOR} of Labeling: Re-evaluating Topic Models for Content Analysis",
|
| 64 |
+
author = "Li, Zongxia and
|
| 65 |
+
Mao, Andrew and
|
| 66 |
+
Stephens, Daniel and
|
| 67 |
+
Goel, Pranav and
|
| 68 |
+
Walpole, Emily and
|
| 69 |
+
Dima, Alden and
|
| 70 |
+
Fung, Juan and
|
| 71 |
+
Boyd-Graber, Jordan",
|
| 72 |
+
editor = "Graham, Yvette and
|
| 73 |
+
Purver, Matthew",
|
| 74 |
+
booktitle = "Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)",
|
| 75 |
+
month = mar,
|
| 76 |
+
year = "2024",
|
| 77 |
+
address = "St. Julian{'}s, Malta",
|
| 78 |
+
publisher = "Association for Computational Linguistics",
|
| 79 |
+
url = "https://aclanthology.org/2024.eacl-long.51/",
|
| 80 |
+
pages = "840--859"
|
| 81 |
+
}
|
| 82 |
+
```
|
| 83 |
+
|
| 84 |
+
If you want to use the claim coherence does not generalize to neural topic models:
|
| 85 |
+
```
|
| 86 |
+
@inproceedings{hoyle-etal-2021-automated,
|
| 87 |
+
title = "Is Automated Topic Evaluation Broken? The Incoherence of Coherence",
|
| 88 |
+
author = "Hoyle, Alexander Miserlis and
|
| 89 |
+
Goel, Pranav and
|
| 90 |
+
Hian-Cheong, Andrew and
|
| 91 |
+
Peskov, Denis and
|
| 92 |
+
Boyd-Graber, Jordan and
|
| 93 |
+
Resnik, Philip",
|
| 94 |
+
booktitle = "Advances in Neural Information Processing Systems",
|
| 95 |
+
year = "2021",
|
| 96 |
+
url = "https://arxiv.org/abs/2107.02173",
|
| 97 |
+
}
|
| 98 |
+
```
|
| 99 |
+
|
| 100 |
+
|
| 101 |
+
If you evaluate ground-truth evaluations or stability:
|
| 102 |
+
```
|
| 103 |
+
@inproceedings{hoyle-etal-2022-neural,
|
| 104 |
+
title = "Are Neural Topic Models Broken?",
|
| 105 |
+
author = "Hoyle, Alexander Miserlis and
|
| 106 |
+
Goel, Pranav and
|
| 107 |
+
Sarkar, Rupak and
|
| 108 |
+
Resnik, Philip",
|
| 109 |
+
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
|
| 110 |
+
year = "2022",
|
| 111 |
+
publisher = "Association for Computational Linguistics",
|
| 112 |
+
url = "https://aclanthology.org/2022.findings-emnlp.390",
|
| 113 |
+
doi = "10.18653/v1/2022.findings-emnlp.390",
|
| 114 |
+
pages = "5321--5344",
|
| 115 |
+
}
|
| 116 |
+
```
|