EntityIdentification

What is EntityIdentification?

EntityIdentification is an MCP (Model Context Protocol) server that determines whether two sets of JSON data originate from the same real-world entity. It compares values using exact matching, text normalization, and a generative language model for semantic similarity.

How to use EntityIdentification?

Install the required dependency with pip install genai. Use the provided functions normalize_text, compare_values, and compare_json to compare two JSON objects. The server integrates a language model to assess semantic similarity and returns a final judgment.

Key features of EntityIdentification

• Text normalization (lowercase, punctuation removal, whitespace normalization)
• Exact and semantic value comparison
• Ignoring order when comparing list values
• Key-by-key JSON traversal and comparison
• Generative language model for final entity matching

Use cases of EntityIdentification

• Deduplicating customer records from different databases
• Verifying if two form submissions belong to the same person
• Cleaning merged datasets by detecting duplicate entries
• Checking identity consistency across multiple data sources

FAQ from EntityIdentification

Which language model does EntityIdentification use?

The example uses genai.GenerativeModel("gemini-2.0-flash-thinking-exp") to generate the final matching result.

How do I install EntityIdentification?

Install the required package with pip install genai. No other setup is mentioned.

Does EntityComparison ignore list order?

Yes. When comparing list values, the server ignores element order for semantic comparison.

What output does the server provide?

After comparing JSON objects key-by-key, it uses a language model to produce a textual judgment on whether the two data sets come from the same entity.

Does EntityIdentification have any known limitations?

The README does not list specific limitations. It only describes the comparison logic and example usage with a single language model.