Added details to README
Browse files
README.md
CHANGED
@@ -1,2 +1,61 @@
|
|
1 |
# CoronaCentral BERT Model for Topic / Article Type Classification
|
2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
# CoronaCentral BERT Model for Topic / Article Type Classification
|
2 |
|
3 |
+
This is the topic / article type classification for the [CoronaCentral website](https://coronacentral.ai). This forms part of the pipeline for downloading and processing coronavirus literature described in the [corona-ml repo](https://github.com/jakelever/corona-ml) with available [step-by-step descriptions](https://github.com/jakelever/corona-ml/blob/master/stepByStep.md). The method is described in the [preprint](https://doi.org/10.1101/2020.12.21.423860) and detailed performance results can be found in the [machine learning details](https://github.com/jakelever/corona-ml/blob/master/machineLearningDetails.md) document.
|
4 |
+
|
5 |
+
This is derived from the [microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract) model and fine-tuned for the sequence classification task.
|
6 |
+
|
7 |
+
## Usage
|
8 |
+
|
9 |
+
Below are two Google Colab notebooks with example usage of this sequence classification model using HuggingFace transformers and KTrain.
|
10 |
+
|
11 |
+
- [HuggingFace example on Google Colab](https://colab.research.google.com/drive/1cBNgKd4o6FNWwjKXXQQsC_SaX1kOXDa4?usp=sharing)
|
12 |
+
- [KTrain example on Google Colab](https://colab.research.google.com/drive/1h7oJa2NDjnBEoox0D5vwXrxiCHj3B1kU?usp=sharing)
|
13 |
+
|
14 |
+
## Training Data
|
15 |
+
|
16 |
+
The model is trained on ~3200 manually-curated articles sampled at various stages during the coronavirus pandemic. The code for training is available in the [category\_prediction](https://github.com/jakelever/corona-ml/tree/master/category_prediction) directory of the main Github Repo. The data is available in the [annotated_documents.json.gz](https://github.com/jakelever/corona-ml/blob/master/category_prediction/annotated_documents.json.gz) file.
|
17 |
+
|
18 |
+
## Inputs and Outputs
|
19 |
+
|
20 |
+
The model takes in a tokenized title and abstract (combined into a single string and separated by a new line). The outputs are topics and article types, broadly called categories in the pipeline code. The types are listed below. Some others are managed by hand-coded rules described in the [step-by-step descriptions](https://github.com/jakelever/corona-ml/blob/master/stepByStep.md).
|
21 |
+
|
22 |
+
### List of Article Types
|
23 |
+
|
24 |
+
- Comment/Editorial
|
25 |
+
- Meta-analysis
|
26 |
+
- News
|
27 |
+
- Review
|
28 |
+
|
29 |
+
### List of Topics
|
30 |
+
|
31 |
+
- Clinical Reports
|
32 |
+
- Communication
|
33 |
+
- Contact Tracing
|
34 |
+
- Diagnostics
|
35 |
+
- Drug Targets
|
36 |
+
- Education
|
37 |
+
- Effect on Medical Specialties
|
38 |
+
- Forecasting & Modelling
|
39 |
+
- Health Policy
|
40 |
+
- Healthcare Workers
|
41 |
+
- Imaging
|
42 |
+
- Immunology
|
43 |
+
- Inequality
|
44 |
+
- Infection Reports
|
45 |
+
- Long Haul
|
46 |
+
- Medical Devices
|
47 |
+
- Misinformation
|
48 |
+
- Model Systems & Tools
|
49 |
+
- Molecular Biology
|
50 |
+
- Non-human
|
51 |
+
- Non-medical
|
52 |
+
- Pediatrics
|
53 |
+
- Prevalence
|
54 |
+
- Prevention
|
55 |
+
- Psychology
|
56 |
+
- Recommendations
|
57 |
+
- Risk Factors
|
58 |
+
- Surveillance
|
59 |
+
- Therapeutics
|
60 |
+
- Transmission
|
61 |
+
- Vaccines
|