Copyright Stats Extractor

A lightweight utility package that parses news headlines or short articles about digital copyright enforcement and automatically extracts key statistics such as the number of takedown requests processed, the year, and the entities involved.
The extractor uses a large language model (LLM) under the hood; by default it uses ChatLLM7 from the langchain_llm7 package, but you can plug in any LangChain chat model you prefer.

📦 Installation

pip install copyright_stats_extractor

🚀 Getting Started

from copyright_stats_extractor import copyright_stats_extractor

# Example text to analyse
user_input = """
In 2023, the Digital Society Agency issued 12,000 takedown requests against
unauthorized streaming sites. Major platforms such as StreamTop and IndiePlay
reported compliance with 95% of the requests. These actions were part of
the global crackdown on digital piracy led by the International Digital
Rights Alliance (IDRA).
"""

# Use the default LLM7 implementation
stats = copyright_stats_extractor(user_input)

print(stats)

Output

[
  "year: 2023",
  "takedown_requests: 12,000",
  "platforms_involved: StreamTop, IndiePlay",
  "authority: International Digital Rights Alliance (IDRA)"
]

🔌 Using a Custom LLM

You can provide any LangChain chat model. Examples:

OpenAI

from langchain_openai import ChatOpenAI
from copyright_stats_extractor import copyright_stats_extractor

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0.2)
stats = copyright_stats_extractor(user_input, llm=llm)

Anthropic

from langchain_anthropic import ChatAnthropic
from copyright_stats_extractor import copyright_stats_extractor

llm = ChatAnthropic(model="claude-3-5-sonnet-20240620", temperature=0.2)
stats = copyright_stats_extractor(user_input, llm=llm)

Google Gemini

from langchain_google_genai import ChatGoogleGenerativeAI
from copyright_stats_extractor import copyright_stats_extractor

llm = ChatGoogleGenerativeAI(model="gemini-1.5-pro", temperature=0.2)
stats = copyright_stats_extractor(user_input, llm=llm)

Note: Any LangChain-compliant chat model can be supplied via the llm argument.

⚙️ Configuration

The default ChatLLM7 uses the free tier which is more than adequate for most use cases. For higher throughput, supply a personal API key:

export LLM7_API_KEY="your_api_key_here"

or pass it directly:

stats = copyright_stats_extractor(user_input, api_key="your_api_key_here")

You can obtain a free API key by registering at https://token.llm7.io/.

📄 Documentation of Output

The function returns a list of strings, each string containing a key‑value pair extracted from the input. The keys correspond to the statistics recognized by the model (e.g. year, takedown_requests, platforms_involved, authority). The format of each string is controlled by an internal prompt that enforces a regular‑expression pattern. If you need a different output structure, customize the prompt and the regex accordingly.

📈 Limitations

• The extraction accuracy depends on the quality of the LLM prompt and the input text length.
• The default free tier for ChatLLM7 may impose request limits; if you hit them, upgrade your API key.

🐛 Issues

Please file bugs or feature requests at the GitHub issues tracker:

https://github.com/chigwell/copyright-stats-extractor/issues

📢 Author

• Eugene Evstafev
  Email: hi@euegne.plus
  GitHub: https://github.com/chigwell