DrishtiSharma's picture
Update app.py
e646334 verified
raw
history blame
15.9 kB
# to-do: add inventor + assignee
import streamlit as st
from crewai import Agent, Task, Crew
import os
from langchain_groq import ChatGroq
from langchain_openai import ChatOpenAI
from fpdf import FPDF
import pandas as pd
import plotly.express as px
import tempfile
import time
import logging
# Setup logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
# Title and Application Introduction
st.title("Patent Strategy and Innovation Consultant")
st.sidebar.write(
"This application uses AI to provide actionable insights and comprehensive analysis for patent-related strategies."
)
# User Input Section
st.sidebar.header("User Inputs")
patent_area = st.text_input("Enter Patent Technology Area", value="Transparent Antennas for Windshields")
stakeholder = st.text_input("Enter Stakeholder", value="Patent Attorneys")
# Initialize LLM
llm = None
# Model Selection
model_choice = st.radio("Select LLM", ["GPT-4o", "llama-3.3-70b"], index=1, horizontal=True)
# API Key Validation and LLM Initialization
groq_api_key = os.getenv("GROQ_API_KEY")
openai_api_key = os.getenv("OPENAI_API_KEY")
#llm = ChatGroq(groq_api_key=os.getenv("GROQ_API_KEY"), model="groq/llama-3.3-70b-versatile")
if model_choice == "llama-3.3-70b":
if not groq_api_key:
st.error("Groq API key is missing. Please set the GROQ_API_KEY environment variable.")
llm = None
else:
llm = ChatGroq(groq_api_key=groq_api_key, model="groq/llama-3.3-70b-versatile")
elif model_choice == "GPT-4o":
if not openai_api_key:
st.error("OpenAI API key is missing. Please set the OPENAI_API_KEY environment variable.")
llm = None
else:
llm = ChatOpenAI(api_key=openai_api_key, model="gpt-4o")
# Advanced Options
st.sidebar.header("Advanced Options")
enable_advanced_analysis = st.sidebar.checkbox("Enable Advanced Analysis", value=True)
enable_custom_visualization = st.sidebar.checkbox("Enable Custom Visualizations", value=True)
# Agent Customization
st.sidebar.header("Agent Customization")
with st.sidebar.expander("Customize Agent Goals", expanded=False):
enable_customization = st.checkbox("Enable Custom Goals")
if enable_customization:
planner_goal = st.text_area(
"Planner Goal",
value="Research trends in patent filings and technological innovation, identify key players, and provide strategic recommendations."
)
writer_goal = st.text_area(
"Writer Goal",
value="Craft a professional insights document summarizing trends, strategies, and actionable outcomes for stakeholders."
)
analyst_goal = st.text_area(
"Analyst Goal",
value=(
"Perform detailed statistical analysis of patent filings, growth trends, and innovation distribution. "
"Identify top assignees, prominent inventors, and key companies in the transparent antenna industry. "
"Provide structured output in a list of dictionaries with 'Category' and 'Values' keys for clear data presentation."
)
)
else:
planner_goal = "Research trends in patent filings and technological innovation, identify key players, and provide strategic recommendations."
writer_goal = "Craft a professional insights document summarizing trends, strategies, and actionable outcomes for stakeholders."
analyst_goal = (
"Perform detailed statistical analysis of patent filings, growth trends, and innovation distribution. "
"Identify top assignees, prominent inventors, and key companies in the transparent antenna industry. "
"Provide structured output in a list of dictionaries with 'Category' and 'Values' keys for clear data presentation."
)
# Agent Definitions
planner = Agent(
role="Patent Research Consultant",
goal=planner_goal,
backstory=(
"You're tasked with researching {topic} patents and identifying key trends and players. Your work supports the Patent Writer and Data Analyst."
),
allow_delegation=False,
verbose=True,
llm=llm
)
writer = Agent(
role="Patent Insights Writer",
goal=writer_goal,
backstory=(
"Using the research from the Planner and data from the Analyst, craft a professional document summarizing patent insights for {stakeholder}."
),
allow_delegation=False,
verbose=True,
llm=llm
)
analyst = Agent(
role="Patent Data Analyst",
goal=analyst_goal,
backstory=(
"Analyze patent filing data and innovation trends in {topic} to provide statistical insights. Your analysis will guide the Writer's final report."
),
allow_delegation=False,
verbose=True,
llm=llm
)
# Task Definitions
plan = Task(
description=(
"1. Research recent trends in {topic} patent filings and innovation.\n"
"2. Identify key players and emerging technologies.\n"
"3. Provide recommendations for stakeholders on strategic directions.\n"
"4. Identify key statistics such as top regions, top players, and hot areas of innovation.\n"
"5. Limit the output to 500 words."
),
expected_output="A research document with structured insights, strategic recommendations, and key statistics.",
agent=planner
)
write = Task(
description=(
"1. Use the Planner's and Analyst's outputs to craft a professional patent insights document.\n"
"2. Include key findings, visual aids, and actionable strategies.\n"
"3. Suggest strategic directions and highlight untapped innovation areas.\n"
"4. Incorporate summarized tables for key statistics and example inventions.\n"
"5. Limit the document to 600 words."
),
expected_output="A polished, stakeholder-ready patent insights document with actionable recommendations.",
agent=writer
)
analyse = Task(
description=(
"1. Conduct a comprehensive statistical analysis of patent filing trends, innovation hot spots, and future growth projections in the transparent antenna industry.\n"
"2. Identify and rank the top regions, leading assignees, prominent inventors, and key companies driving innovation.\n"
"3. Highlight regional innovation trends and the distribution of emerging technologies across different geographies.\n"
"4. Provide actionable insights and strategic recommendations based on the data.\n"
"5. Deliver structured output in a list of dictionaries with 'Category' and 'Values' fields:\n"
" - 'Values' can be:\n"
" a) A dictionary with counts for quantitative data (e.g., {'Region A': 120, 'Region B': 95}),\n"
" b) A list of key items (technologies, companies, inventors), or\n"
" c) Descriptive text for qualitative insights.\n"
"6. Example Output Format:\n"
"[\n"
" {'Category': 'Top Regions', 'Values': {'North America': 120, 'Europe': 95, 'Asia-Pacific': 85}},\n"
" {'Category': 'Top Assignees', 'Values': {'Company A': 40, 'Company B': 35}},\n"
" {'Category': 'Top Inventors', 'Values': {'Inventor X': 25, 'Inventor Y': 20}},\n"
" {'Category': 'Emerging Technologies', 'Values': ['Graphene Antennas', '5G Integration']},\n"
" {'Category': 'Strategic Insights', 'Values': 'Collaborations between automotive and material science industries are accelerating innovation.'}\n"
"]\n"
"7. Ensure that the output is clean, well-structured, and formatted for use in visualizations and tables."
),
expected_output="A structured, well-organized dataset with numeric, list-based, and descriptive insights for comprehensive visual and tabular reporting.",
agent=analyst
)
crew = Crew(
agents=[planner, analyst, writer],
tasks=[plan, analyse, write],
verbose=True
)
# PDF Report Generation
def generate_pdf_report(result, charts=None, table_data=None, metadata=None):
with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_pdf:
pdf = FPDF()
pdf.add_page()
pdf.set_font("Arial", size=12)
pdf.set_auto_page_break(auto=True, margin=15)
pdf.set_font("Arial", size=16, style="B")
pdf.cell(200, 10, txt="Patent Strategy and Innovation Report", ln=True, align="C")
pdf.ln(10)
if metadata:
pdf.set_font("Arial", size=10)
for key, value in metadata.items():
pdf.cell(200, 10, txt=f"{key}: {value}", ln=True)
pdf.set_font("Arial", size=12)
pdf.multi_cell(0, 10, txt=result)
if charts:
for chart_path in charts:
try:
pdf.add_page()
pdf.image(chart_path, x=10, y=20, w=180)
logging.info(f"Successfully included chart: {chart_path}")
except Exception as e:
logging.error(f"Failed to include chart in PDF: {chart_path}. Error: {e}")
if table_data:
pdf.add_page()
pdf.set_font("Arial", size=10)
pdf.cell(200, 10, txt="Consolidated Table:", ln=True, align="L")
for row in table_data:
pdf.cell(200, 10, txt=str(row), ln=True)
pdf.output(temp_pdf.name)
return temp_pdf.name
# Data Validation
def validate_analyst_output(analyst_output):
if not analyst_output:
st.warning("No data available for analysis.")
return None
if not isinstance(analyst_output, list) or not all(isinstance(item, dict) for item in analyst_output):
st.warning("Analyst output must be a list of dictionaries.")
return None
required_keys = {'Category', 'Values'}
if not all(required_keys.issubset(item.keys()) for item in analyst_output):
st.warning(f"Each dictionary must contain keys: {required_keys}")
return None
return analyst_output
# Visualization and Table Display
def create_visualizations(analyst_output):
chart_paths = []
validated_data = validate_analyst_output(analyst_output)
if validated_data:
for item in validated_data:
category = item["Category"]
values = item["Values"]
# Handle dictionary-type data for bar charts
if isinstance(values, dict):
df = pd.DataFrame(list(values.items()), columns=["Label", "Count"])
chart = px.bar(df, x="Label", y="Count", title=f"{category} Analysis")
# Handle list-type data for pie charts
elif isinstance(values, list):
df = pd.DataFrame(values, columns=["Label"])
df["Count"] = 1 # Assign count for visualization
chart = px.pie(df, names="Label", values="Count", title=f"{category} Distribution")
# Handle text data by converting it into a simple table
elif isinstance(values, str):
st.subheader(f"{category} Insights")
st.table(pd.DataFrame({"Insights": [values]}))
continue # Skip visualization for text data
else:
st.warning(f"Unsupported data format in category: {category}")
continue
# Display the chart in Streamlit
st.plotly_chart(chart)
# Save the chart for PDF generation
with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as temp_chart:
chart.write_image(temp_chart.name)
chart_paths.append(temp_chart.name)
return chart_paths
def display_table(analyst_output):
table_data = []
validated_data = validate_analyst_output(analyst_output)
if validated_data:
for item in validated_data:
category = item["Category"]
values = item["Values"]
try:
# Dictionary → Structured Table
if isinstance(values, dict):
df = pd.DataFrame(list(values.items()), columns=["Label", "Count"])
st.subheader(f"{category} (Table View)")
st.dataframe(df)
table_data.extend(df.to_dict(orient="records"))
# List → Simple List Table
elif isinstance(values, list):
df = pd.DataFrame(values, columns=["Items"])
st.subheader(f"{category} (List View)")
st.dataframe(df)
table_data.extend(df.to_dict(orient="records"))
# String → Text Table
elif isinstance(values, str):
st.subheader(f"{category} (Summary)")
st.table(pd.DataFrame({"Insights": [values]}))
table_data.append({"Category": category, "Values": values})
else:
st.warning(f"Unsupported data format for category: {category}")
except Exception as e:
logging.error(f"Error displaying {category} as table: {e}")
st.error(f"Failed to display {category} as a table.")
return table_data
def parse_analyst_output(raw_output):
structured_data = []
current_category = None
lines = raw_output.split('\n')
for line in lines:
line = line.strip()
if line.startswith("Category:"):
current_category = line.replace("Category:", "").strip()
elif line.startswith("Values:"):
continue
elif line and current_category:
structured_data.append({
"Category": current_category,
"Values": line
})
return structured_data
# Main Execution Block
if st.button("Generate Patent Insights"):
with st.spinner('Processing...'):
try:
start_time = time.time()
results = crew.kickoff(inputs={"topic": patent_area, "stakeholder": stakeholder})
elapsed_time = time.time() - start_time
writer_output = getattr(results.tasks_output[2], "raw", "No details available.")
if writer_output:
st.markdown("### Final Report")
st.write(writer_output)
else:
st.warning("No final report available.")
with st.expander("Explore Detailed Insights"):
tab1, tab2 = st.tabs(["Planner's Insights", "Analyst's Analysis"])
with tab1:
planner_output = getattr(results.tasks_output[0], "raw", "No details available.")
st.write(planner_output)
with tab2:
analyst_output = getattr(results.tasks_output[1], "raw", "No details available.")
st.write(analyst_output)
# Convert raw text to structured data
if isinstance(analyst_output, str):
analyst_output = parse_analyst_output(analyst_output)
st.subheader("Structured Analyst Output")
st.write(analyst_output)
charts = []
if enable_advanced_analysis:
charts = create_visualizations(analyst_output)
table_data = display_table(analyst_output)
st.success(f"Analysis completed in {elapsed_time:.2f} seconds.")
pdf_path = generate_pdf_report(writer_output, charts=charts, table_data=table_data, metadata={"Technology Area": patent_area, "Stakeholder": stakeholder})
with open(pdf_path, "rb") as report_file:
st.download_button("Download Report", data=report_file, file_name="Patent_Strategy_Report.pdf")
except Exception as e:
logging.error(f"An error occurred during execution: {e}")
st.error(f"An error occurred during execution: {e}")