Commit
•
c6097b0
1
Parent(s):
45e060f
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import pandas as pd
|
3 |
+
import numpy as np
|
4 |
+
|
5 |
+
# Global variable to store the DataFrame
|
6 |
+
df = None
|
7 |
+
# Global variable to keep track of the current row index
|
8 |
+
current_row = 0
|
9 |
+
|
10 |
+
def load_csv(file):
|
11 |
+
global df
|
12 |
+
global current_row
|
13 |
+
# import the csv and set the data types to be int, string, string, string, string, string, string
|
14 |
+
df = pd.read_csv(file.name, dtype={'id':int, 'hs': str, 'cs': str, 'topic': str, 'tone': str, 'isCSContextuallyRelevant': str, 'isToneMatch': str})
|
15 |
+
current_row = 0
|
16 |
+
row_dict = df.iloc[current_row].to_dict()
|
17 |
+
return row_dict['id'], row_dict['hs'], row_dict['cs'], row_dict['topic'], row_dict['tone'], row_dict['isCSContextuallyRelevant'], row_dict['isToneMatch']
|
18 |
+
|
19 |
+
def annotate_row(isCSContextuallyRelevant, isToneMatch):
|
20 |
+
global df
|
21 |
+
global current_row
|
22 |
+
|
23 |
+
df.at[current_row, 'isCSContextuallyRelevant'] = isCSContextuallyRelevant
|
24 |
+
df.at[current_row, 'isToneMatch'] = isToneMatch
|
25 |
+
|
26 |
+
if current_row < len(df) - 1:
|
27 |
+
current_row += 1
|
28 |
+
else:
|
29 |
+
current_row = 0
|
30 |
+
df.to_csv('annotated_data.csv', index=False)
|
31 |
+
|
32 |
+
row_dict = df.iloc[current_row].to_dict()
|
33 |
+
return row_dict['id'], row_dict['hs'], row_dict['cs'], row_dict['topic'], row_dict['tone'], row_dict['isCSContextuallyRelevant'], row_dict['isToneMatch'], 'annotated_data.csv'
|
34 |
+
|
35 |
+
def navigate(direction):
|
36 |
+
global current_row
|
37 |
+
if direction == "Previous":
|
38 |
+
current_row = max(0, current_row - 1)
|
39 |
+
elif direction == "Next":
|
40 |
+
current_row = min(len(df) - 1, current_row + 1)
|
41 |
+
elif direction == "First Unlabeled":
|
42 |
+
unlabeled_row = df[df['isCSContextuallyRelevant'].isna()].index.min()
|
43 |
+
if not np.isnan(unlabeled_row):
|
44 |
+
current_row = int(unlabeled_row)
|
45 |
+
|
46 |
+
row_dict = df.iloc[current_row].to_dict()
|
47 |
+
return row_dict['id'], row_dict['hs'], row_dict['cs'], row_dict['topic'], row_dict['tone'], row_dict['isCSContextuallyRelevant'], row_dict['isToneMatch']
|
48 |
+
|
49 |
+
with gr.Blocks(theme=gr.themes.Soft()) as annotator:
|
50 |
+
gr.Markdown("## Data Annotation")
|
51 |
+
|
52 |
+
with gr.Row():
|
53 |
+
gr.Markdown("### Upload CSV")
|
54 |
+
file_upload = gr.File()
|
55 |
+
btn_load = gr.Button("Load CSV")
|
56 |
+
|
57 |
+
with gr.Row():
|
58 |
+
gr.Markdown("### Current Row")
|
59 |
+
with gr.Row():
|
60 |
+
idx = gr.Number(label='Index')
|
61 |
+
hs = gr.Textbox(label='HS')
|
62 |
+
cs = gr.Textbox(label='CS')
|
63 |
+
|
64 |
+
with gr.Row():
|
65 |
+
topic = gr.Textbox(label='Topic')
|
66 |
+
tone = gr.Textbox(label='Tone')
|
67 |
+
|
68 |
+
with gr.Row():
|
69 |
+
isCSContextuallyRelevant = gr.Radio(["1", "0"], label="Contextually Relevant?")
|
70 |
+
isToneMatch = gr.Radio(["1", "0"], label="Tone Match?")
|
71 |
+
btn_annotate = gr.Button("Annotate")
|
72 |
+
|
73 |
+
with gr.Row():
|
74 |
+
btn_previous = gr.Button("Previous")
|
75 |
+
btn_next = gr.Button("Next")
|
76 |
+
btn_first_unlabeled = gr.Button("First Unlabeled")
|
77 |
+
|
78 |
+
with gr.Row():
|
79 |
+
gr.Markdown("### Annotated Data File Download")
|
80 |
+
file_download = gr.File()
|
81 |
+
|
82 |
+
btn_load.click(load_csv, inputs=[file_upload], outputs=[idx, hs, cs, topic, tone, isCSContextuallyRelevant, isToneMatch], scroll_to_output=True)
|
83 |
+
btn_annotate.click(annotate_row, inputs=[isCSContextuallyRelevant, isToneMatch], outputs=[idx, hs, cs, topic, tone, isCSContextuallyRelevant, isToneMatch, file_download])
|
84 |
+
btn_previous.click(navigate, inputs=gr.Textbox("Previous", visible=False), outputs=[idx, hs, cs, topic, tone, isCSContextuallyRelevant, isToneMatch])
|
85 |
+
btn_next.click(navigate, inputs=gr.Textbox("Next", visible=False), outputs=[idx, hs, cs, topic, tone, isCSContextuallyRelevant, isToneMatch])
|
86 |
+
btn_first_unlabeled.click(navigate, inputs=gr.Textbox("First Unlabeled", visible=False), outputs=[idx, hs, cs, topic, tone, isCSContextuallyRelevant, isToneMatch])
|
87 |
+
|
88 |
+
annotator.launch()
|