European VAT Rule Analysis Assistant

How is the VAT rule data indexed and embeddings created?

• VAT rule data is sourced from various European VAT authorities, including national tax authorities and official VAT databases.
• The data is collected and standardized to ensure consistency and accuracy.
• Textual information from the VAT rules, such as legal texts and guidelines, is extracted and preprocessed. Special characters and irrelevant elements are removed, and the text is tokenized.
• Data is then indexed using a custom indexing algorithm designed for VAT rules, which takes into account the structure and hierarchies of the regulations.
• Embeddings are created using the OpenAI text-embedding-ada-002 model, producing high-dimensional vectors with a dimension of 1536.
• Vector embeddings are stored in a dedicated database optimized for high-dimensional data storage and retrieval.

How do we keep the data updated?

Our system maintains connections with various European VAT authorities to ensure access to the latest VAT rule updates. We regularly monitor changes and updates in VAT regulations and collect the revised documents or guidelines. The new data is processed, indexed, and embeddings are generated, ensuring that our AI embeddings reflect the most up-to-date VAT rule information.

How to use

• Python
• Curl

Python Code

You can access the VAT rule embeddings using the embedelite Python package or by making a CURL request.

# This returns a list of embeddings
from embedelite import load_embedding

embeddings = load_embedding("vat-rules-eu")
print(len(embeddings))
print(embeddings[1])

# This returns an object that can be directly inserted into the database
result = load_embedding("vat-rules-eu", embed_for="qdrant")
# The result is {"embeddings": [], "documents": [], "ids": []}
print(result["embeddings"])
print(result["documents"])
print(result["ids"])

Curl

Request

curl -X POST -H "Content-Type: application/json" -d '{
  "doc_id": "vat-rules-eu"
}' https://api.embedelite.com/v1/embeddings/download/

Response

{
  "mappings": {
    "properties": {
      "doc_source": { "type": "keyword" },
      "sentence": { "type": "text" },
      "embeddings": { "type": "dense_vector", "dims": 1536, "index": false },
      "doc_source": { "type": "keyword" }
    }
  }
}

Data credit to various European VAT authorities, without which this embedding wouldn’t be possible.