12. Upstage Layout Analysis Loader

UpstageLayoutAnalysisLoader Is a document analysis tool provided by Upstage AI, a document loader that can be used in conjunction with the LangChain framework.

Main features: -Perform layout analysis in various types of documents, including PDFs and images -Automatically recognize and extract structural elements of documents (titles, paragraphs, tables, images, etc.) -OCR function support (optional)

UpstageLayoutAnalysisLoader goes beyond simple text extraction to understand the structure of documents and identify relationships between elements, enabling more accurate document analysis.

install

langchain-upstage Use the package after installation.

Copy

pip install -U langchain-upstage

API Key Settings

.env To file UPSTAGE_API_KEY Set the key.

Reference

• Upstage developer documentation See.

Preferences

Copy

# Configuration file for managing API KEY as environment variable

# Load API KEY information
load_dotenv()

Copy

True

Copy

# LangSmith Set up tracking. https://smith.langchain.com
# !pip install langchain-teddynote
from langchain_teddynote import logging

# Enter a project name.
logging.langsmith("CH07-DocumentLoader")

Copy

 Start tracking LangSmith. 
[Project name] 
CH07-DocumentLoader 

UpstageLayoutAnalysisLoader

Main parameters

• file_path : Document path to analyze

• output_type : Output format [(default)'html','text']

• split : Document splitting method ['none','element','page']

• use_ocr=True : Using OCR

• exclude=["header", "footer"] : Header, except Footer

Copy

from langchain_upstage import UpstageLayoutAnalysisLoader

# File Path
file_path = "./data/SPRI_AI_Brief_2023년12월호_F.pdf"

# Document loader settings
loader = UpstageLayoutAnalysisLoader(
    file_path,
    output_type="text",
    split="page",
    use_ocr=True,
    exclude=["header", "footer"],
)

# Load Document
docs = loader.load()

# Output the results
for doc in docs[:3]:
    print(doc)

Copy

page_content='SPRi AI Brief The latest trends in the AI industry December 2023' metadata={'page': 1} 
page_content='December 2023 CONTENTS I Artificial Industry Trend Brief 1. Policy/Act ▷ United States announces administrative order on safe and reliable AI development and use 1 
▷ G7, Hiroshima AI process to agree on international action decree for AI companies 2 
▷ 28 countries participating in the UK AI Safety Summit, Joint Response to AI Risk 3 
▷ US court dismisses copyright lawsuits filed by artists to AI companies 4 
▷ US Federal Trade Commission submits AI comments on consumer protection and competition to Copyright Office 5 
▷ EU AI law third-party negotiations, based model regulation views on turbulence 6 2. Corporate/Industrial ▷ American Frontier Model Forum raises $10 million AI safety fund 7 
▷ Cohir unveils data source explorer to ensure data transparency 8 
▷ Alibaba Cloud unveils the latest LLM'Gunichien One 2.0' 9 
▷ Samsung Electronics unveils self-developed AI'Samsung Gauss' 10 
▷ Google strengthens AI cooperation generated by $20 billion investment in Antwerp 11 
▷ IDC forecasts $250 billion in AI software sales in 2027 12 
▷ Bill Gates predicts paradigm shift in computer use due to AI agents 13 
▷ YouTube mandates AI-generated content display from 2024 14 3. Technology / Research ▷ UK Science and Innovation Technology Department Announces Establishment of AI Safety Institute 15 
▷ Google Deepmind announces classification system for features and behavior of universal AI models 16 
▷ GPT-4 is the best in Galileo's LLM hallucinations index assessment 17 4. Workforce/Education ▷ Oxford Internet Institute, UK, AI technician wages average 21% higher 18 II. Main Events ▷ CES 2024 19 
▷ AIMLA 2024 19 
▷ AAAI Conference on Artificial Intelligence 19' metadata={'page': 2} 
page_content='I. Artificial Industry Trend Brief' metadata={'page': 3}