#!/usr/bin/env python
# coding: utf-8
# In[ ]:
import gradio as gr
import pandas as pd
from transformers import TapasTokenizer, TapasForQuestionAnswering
from transformers import AutoTokenizer, AutoModelForTableQuestionAnswering
from transformers import pipeline
import PIL
# read the module_guide_tableQA\0915NC_Studienplaetze.jpg as pil image
pil_image = PIL.Image.open("0915NC_Studienplaetze.jpg")
# make that image a bit less high
pil_image = pil_image.resize((int(pil_image.width * 0.5), int(pil_image.height * 0.5)))
# In[ ]:
def get_answer(
dropdown,
question,
view_as_table=False,
model="google/tapas-finetuned-wtq",
#progress=gr.Progress(),
):
#progress(0, desc="Looking for answer in module guide...")
df = pd.DataFrame()
if dropdown == "Master Information Systems":
df = pd.read_excel(
r"03_extracted_final_modules\MS_IS_all_modules_orginal_15_rows_cleaned.xlsx"
)
elif dropdown == "Bachelor Information Systems":
df = pd.read_excel(r"03_extracted_final_modules\BA_IS_all_modules_15.xlsx")
elif dropdown == "Bachelor Management":
df = pd.read_excel(r"03_extracted_final_modules\BA_MM_all_modules_15.xlsx")
df = df.astype(str)
print(question)
question = str(question)
print(df.shape)
question = [question]
if model == "google/tapas-finetuned-wtq":
tqa = pipeline(
task="table-question-answering", model="google/tapas-base-finetuned-wtq"
)
elif model == "google/tapas-large-finetuned-wtq":
tqa = pipeline(
task="table-question-answering", model="google/tapas-large-finetuned-wtq"
)
results = tqa(table=df, query=question)
print(results)
cells_input = results["cells"]
cells_input = str(cells_input)
cells_input = cells_input.replace("[", "")
cells_input = cells_input.replace("]", "")
cells_input = cells_input.replace("'", "")
print(cells_input)
print(results)
html_string_short = f"Short Answer:
{cells_input}
"
row_numbers = [coord[0] for coord in results["coordinates"]]
df_short = df.iloc[row_numbers]
df_short = df_short.dropna(axis=1, how="all")
df_short = df_short.loc[:, (df_short != "--").any(axis=0)]
html_table = (
f"
Complete Module(s):
{df_short.to_html(index=False)}
"
)
# check if there are more than 1 rows in df_short
html_string = ""
if df_short.shape[0] > 1 or view_as_table == True:
html_string = html_table
elif df_short.shape[0] == 1:
html_string = """
Detailed Module Information
Module title:
Project Seminar
Abbreviation:
12-PS-192-m01
Module coordinator:
Faculty of Business Management and Economics
Module offered by:
Holder of the Chair of Business Management and Business
Method of grading:
numerical grade
Contents:
In small project teams of 4 to 10 members, students will spend several months actively working on a specific and realistic problem with practical relevance. They will progress through several project stages including as-is analysis, to-be conception and implementation of an IS solution. The project teams will be required to work independently and will only receive advice and minor support from research assistants.
Intended learning outcomes:
- Analyze business tasks and requirements and generate fitting IS solutions
- Apply project management methods
- Internalize stress, time and conflict management by means of practical teamwork
Courses:
Project: preparing a conceptual design (approx. 150 hours), designing and implementing an approach to solution (approx. 300 hours) as well as presentation (approx. 20 minutes), weighted 1:2:1
Language of assessment: German, English
Creditable for bonus
"""
else:
html_string = ""
return html_string_short, html_string
def change_html_link(dropdown_item):
html_link = ""
if dropdown_item == "Master Information Systems":
html_link = f'View complete pdf here: {dropdown_item}
Ask whatever you want to know about the module guide here. You can ask formality-based and content-based questions.
'
elif dropdown_item == "Bachelor Information Systems":
html_link = f'View complete pdf here: {dropdown_item} Ask whatever you want to know about the module guide here. You can ask formality-based and content-based questions.
'
elif dropdown_item == "Bachelor Management":
html_link = f'View complete pdf here: {dropdown_item} Ask whatever you want to know about the module guide here. You can ask formality-based and content-based questions.
'
return html_link
# In[20]:
with gr.Blocks() as demo:
gr.HTML(
"""
"""
)
gr.HTML(
"Your Module Guide Assistant
"
)
table = gr.Dropdown(
[
"Master Information Systems",
"Bachelor Information Systems",
"Bachelor Management",
],
label="Select Module Guide",
value="Master Information Systems",
)
html_link = gr.HTML(
"""
View complete PDF here: Master Information Systems
Ask whatever you want to know about the module guide here. You can ask formality-based and content-based questions.
"""
)
table.change(change_html_link, table, html_link)
question = gr.Textbox(
label="Question", value="How many ECTS credits does the project seminar have?"
)
with gr.Accordion("Advanced Options", open=False):
with gr.Group():
model_selction = gr.Dropdown(
[
"google/tapas-finetuned-wtq",
"google/tapas-large-finetuned-wtq",
],
label="Select Model",
value="google/tapas-finetuned-wtq",
)
view_as_table_or_text = gr.Checkbox(
label="View detailed information as table", value=False
)
ask_btn = gr.Button("Ask The Assistant")
gr.HTML("
")
inputs = [table, question, view_as_table_or_text, model_selction]
output_question = gr.HTML(label="Answer")
outout_full_module = gr.HTML(label="Detailed Description")
outputs = [output_question, outout_full_module]
ask_btn.click(fn=get_answer, inputs=inputs, outputs=outputs, api_name="greet")
demo.launch(debug=True, share=True)
