stormsidali2001
commited on
Commit
•
17d0414
1
Parent(s):
e9ce464
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,93 @@
|
|
1 |
-
---
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: tf-keras
|
3 |
+
pipeline_tag: text-classification
|
4 |
+
widget:
|
5 |
+
- text: "Climate change is a pressing global issue with far-reaching consequences for ecosystems and human societies."
|
6 |
+
output:
|
7 |
+
- label: Show that the research area is important, problematic, or relevant in some way
|
8 |
+
score: 0.95
|
9 |
+
- label: Introduce and review previous research in the field
|
10 |
+
score: 0.05
|
11 |
+
- text: "Numerous studies have investigated the impact of rising temperatures on marine biodiversity."
|
12 |
+
output:
|
13 |
+
- label: Show that the research area is important, problematic, or relevant in some way
|
14 |
+
score: 0.1
|
15 |
+
- label: Introduce and review previous research in the field
|
16 |
+
score: 0.9
|
17 |
+
- text: "Despite its importance, the specific role of ocean currents in mitigating climate change remains poorly understood."
|
18 |
+
output:
|
19 |
+
- label: Show that the research area is important, problematic, or relevant in some way
|
20 |
+
score: 0.55
|
21 |
+
- label: Introduce and review previous research in the field
|
22 |
+
score: 0.45
|
23 |
+
license: mit
|
24 |
+
datasets:
|
25 |
+
- stormsidali2001/IMRAD-introduction-sentences-moves-sub-moves-dataset
|
26 |
+
language:
|
27 |
+
- en
|
28 |
+
metrics:
|
29 |
+
- f1
|
30 |
+
- accuracy
|
31 |
+
base_model: google/bert-base-cased
|
32 |
+
---
|
33 |
+
|
34 |
+
## IMRaD Introduction Move 0 Sub-move Classifier
|
35 |
+
|
36 |
+
This model is a fine-tuned BERT model specialized in classifying sentences from the "Establishing a Research Territory" (Move 0) section of scientific research paper introductions into their corresponding sub-moves:
|
37 |
+
|
38 |
+
* **Show that the research area is important, problematic, or relevant in some way:** Highlighting the significance, issues, or relevance of the research topic.
|
39 |
+
* **Introduce and review previous research in the field:** Presenting a brief overview of existing work and studies related to the topic.
|
40 |
+
|
41 |
+
**Parent Classifier:**
|
42 |
+
|
43 |
+
This model is designed to be used in conjunction with the main IMRaD Introduction Move Classifier: [https://huggingface.co/stormsidali2001/IMRAD_introduction_moves_classifier](https://huggingface.co/stormsidali2001/IMRAD_introduction_moves_classifier).
|
44 |
+
|
45 |
+
The parent classifier identifies the overall IMRaD move for each sentence. If a sentence is classified as "Establishing a Research Territory" (Move 0), this sub-move classifier can be used to further analyze the specific purpose of that sentence within Move 0.
|
46 |
+
|
47 |
+
## Intended Uses & Limitations
|
48 |
+
|
49 |
+
**Intended Uses:**
|
50 |
+
|
51 |
+
* **Scientific Writing Assistance:** Help researchers and students understand and refine the structure of their "Establishing a Research Territory" section.
|
52 |
+
* **Literature Review Analysis:** Quickly identify how authors establish the context and background in research paper introductions.
|
53 |
+
* **Educational Tool:** Illustrate the different sub-moves used to establish a research territory in scientific writing.
|
54 |
+
|
55 |
+
**Limitations:**
|
56 |
+
|
57 |
+
* **Domain Specificity:** The model was trained on scientific research papers and may not be as accurate on other types of text.
|
58 |
+
* **Accuracy:** While the model has good performance, it is not perfect. Predictions should be carefully reviewed.
|
59 |
+
* **Sentence-Level Classification:** The model classifies individual sentences and does not provide an analysis of the entire "Establishing a Research Territory" section as a whole.
|
60 |
+
|
61 |
+
## Training and Evaluation Data
|
62 |
+
|
63 |
+
This model was trained and evaluated on a subset of the "IMRAD Introduction Sentences Moves & Sub-moves Dataset" available on Hugging Face: [https://huggingface.co/datasets/stormsidali2001/IMRAD-introduction-sentences-moves-sub-moves-dataset](https://huggingface.co/datasets/stormsidali2001/IMRAD-introduction-sentences-moves-sub-moves-dataset)
|
64 |
+
|
65 |
+
The dataset includes sentences specifically from Move 0 of introductions, labeled with their respective sub-moves.
|
66 |
+
|
67 |
+
**Training Details:**
|
68 |
+
|
69 |
+
* **Base Model:** `google/bert-base-cased`
|
70 |
+
* **Implementation:** TensorFlow/Keras
|
71 |
+
* **Evaluation Metrics:** F1 score and accuracy
|
72 |
+
|
73 |
+
## How to Use
|
74 |
+
|
75 |
+
```python
|
76 |
+
from transformers import pipeline
|
77 |
+
|
78 |
+
# Load the parent classifier
|
79 |
+
move_classifier = pipeline("text-classification", model="stormsidali2001/IMRAD_introduction_moves_classifier")
|
80 |
+
|
81 |
+
# Load the sub-move classifier for Move 0
|
82 |
+
submove_classifier_0 = pipeline("text-classification", model="stormsidali2001/IMRAD-introduction-move-zero-sub-moves-classifier")
|
83 |
+
|
84 |
+
sentence = "Electronic cigarettes were introduced into the US market in 2007."
|
85 |
+
|
86 |
+
# First, classify the move
|
87 |
+
move_result = move_classifier(sentence)
|
88 |
+
move = move_result[0]['label']
|
89 |
+
|
90 |
+
if move == "Establishing a Research Territory":
|
91 |
+
# If Move 0, classify the sub-move
|
92 |
+
submove_result = submove_classifier_0(sentence)
|
93 |
+
print(submove_result)
|