Why does your Business need the Multilingual Recommended System?<\/h2>\n\n\n\n
Language diversity can make or break a global user experiences. It ensure that every customer can understood, regardless of their native language.<\/p>\n\n\n\n
This approach boosts engagement, increases conversions, and builds loyalty across different markets.<\/p>\n\n\n\n
Building Multilingual Recommended System<\/h2>\n\n\n\nCollecting products data for Multilingual Recommended System<\/h3>\n\n\n\n\n- Gather all product data, with their title, and text descriptions.<\/li>\n\n\n\n
- Ensure the dataset covers all the languages your target audience speaks.<\/li>\n\n\n\n
- Example:<\/strong> A T-shirt’s description is available in English, Hindi, and Polish, or multiple products in different languages, catering to diverse customers.<\/li>\n<\/ul>\n\n\n\n
Select right embedding model<\/h3>\n\n\n\n\n- Selecting the right embedding model is very important. Model should support multilingual.<\/li>\n\n\n\n
- Multilingual embedding models are trained on multiple language data.<\/li>\n\n\n\n
- Two similar products but have been written in different languages, their multilingual models generate similar context embeddings.<\/li>\n\n\n\n
- Some multilingual embeddings models are openai – text-embedding-3-large, cohere embed-multilingual-light-v3.0, embed-multilingual-v3.0, etc.<\/li>\n<\/ul>\n\n\n\n
Creating products embeddings<\/h3>\n\n\n\n
Example, Generating embeddings from an openai<\/p>\n\n\n\n
from openai import OpenAI\nclient = OpenAI()\n\ndef get_embedding(text, model="text-embedding-ada-002"):\n text = text.replace("\\n", " ")\n return client.embeddings.create(input=[text], model=model).data[0].embedding\n\nproduct_embedding = get_embedding("""\nWireless Noise-Canceling Headphones\n\nExperience superior sound quality with our wireless noise canceling headphones.\nDesigned for comfort,durability and they offer up to 30 hours of battery life and seamless bluetooth connectivity.\n""")<\/pre>\n\n\n\nThe output is a vector representation:<\/p>\n\n\n\n
[\n -0.006929283495992422,\n -0.005336422007530928,\n ... (truncated for brevity)\n]<\/pre>\n\n\n\nSave embeddings in VectorDB<\/h3>\n\n\n\n
Store the generated embeddings in a vector database with metadata such as product ID, language, and category. Popular VectorDB include:<\/p>\n\n\n\n
\n- ChromaDB<\/strong><\/li>\n\n\n\n
- Pinecone<\/strong><\/li>\n\n\n\n
- Milvus<\/strong><\/li>\n<\/ul>\n\n\n\n
Recommending Products<\/h3>\n\n\n\nStep 1: Calculate Similarity<\/h4>\n\n\n\n
When a user views a product, API generates embeddings for its title and description. <\/p>\n\n\n\n
Query the vector database using these embeddings to find similar products based on cosine similarity or other distance metrics.<\/p>\n\n\n\n
Step 2: Generate Results<\/h4>\n\n\n\n
The vector database retrieves products with the most similar embeddings, providing recommendations.<\/p>\n\n\n\n
Even if the recommended products are in different languages, the multilingual model ensures they share the same contextual meaning.<\/p>\n\n\n\n
Benefits of a Multilingual Recommender System<\/h3>\n\n\n\n\n- Seamless User Experience:<\/strong> Users see recommendations in their preferred language.<\/li>\n\n\n\n
- Global Reach:<\/strong> Cater to diverse markets without language limitations.<\/li>\n\n\n\n
- Increased Sales:<\/strong> Personalized suggestions drive higher conversions.<\/li>\n\n\n\n
- Efficient Scaling:<\/strong> Multilingual embeddings reduce the need for separate models per language.<\/li>\n<\/ul>\n\n\n\n
You can also check out the Magento 2 Recommendation System<\/a>. It works with a multilingual embeddings model and supports dynamic models.<\/p>\n\n\n\n![\"multilingual](\"https:\/\/cdnblog.webkul.com\/blog\/wp-content\/uploads\/2024\/11\/multilingualrecommendation-969x1024.webp\")
<\/figure>\n\n\n\nThis is a pair of wireless headphones with the product description in English.<\/p>\n\n\n\n
The recommended products are available in different languages: the first in French, the second in Spanish, and the last in Italian.<\/p>\n\n\n\n
Our recommendation system suggests products even in different languages.<\/p>\n\n\n\n
Conclusion<\/h3>\n\n\n\n
A multilingual recommender system is vital for any e-commerce business targeting global markets.<\/p>\n\n\n\n
By leveraging multilingual embeddings, vector databases, and similarity metrics, you can create an inclusive and effective recommendation engine that resonates with users worldwide.<\/p>\n\n\n\n
- \n
- Gather all product data, with their title, and text descriptions.<\/li>\n\n\n\n
- Ensure the dataset covers all the languages your target audience speaks.<\/li>\n\n\n\n
- Example:<\/strong> A T-shirt’s description is available in English, Hindi, and Polish, or multiple products in different languages, catering to diverse customers.<\/li>\n<\/ul>\n\n\n\n
Select right embedding model<\/h3>\n\n\n\n
- \n
- Selecting the right embedding model is very important. Model should support multilingual.<\/li>\n\n\n\n
- Multilingual embedding models are trained on multiple language data.<\/li>\n\n\n\n
- Two similar products but have been written in different languages, their multilingual models generate similar context embeddings.<\/li>\n\n\n\n
- Some multilingual embeddings models are openai – text-embedding-3-large, cohere embed-multilingual-light-v3.0, embed-multilingual-v3.0, etc.<\/li>\n<\/ul>\n\n\n\n
Creating products embeddings<\/h3>\n\n\n\n
Example, Generating embeddings from an openai<\/p>\n\n\n\n
from openai import OpenAI\nclient = OpenAI()\n\ndef get_embedding(text, model="text-embedding-ada-002"):\n text = text.replace("\\n", " ")\n return client.embeddings.create(input=[text], model=model).data[0].embedding\n\nproduct_embedding = get_embedding("""\nWireless Noise-Canceling Headphones\n\nExperience superior sound quality with our wireless noise canceling headphones.\nDesigned for comfort,durability and they offer up to 30 hours of battery life and seamless bluetooth connectivity.\n""")<\/pre>\n\n\n\nThe output is a vector representation:<\/p>\n\n\n\n
[\n -0.006929283495992422,\n -0.005336422007530928,\n ... (truncated for brevity)\n]<\/pre>\n\n\n\n
Save embeddings in VectorDB<\/h3>\n\n\n\n
Store the generated embeddings in a vector database with metadata such as product ID, language, and category. Popular VectorDB include:<\/p>\n\n\n\n
- \n
- ChromaDB<\/strong><\/li>\n\n\n\n
- Pinecone<\/strong><\/li>\n\n\n\n
- Milvus<\/strong><\/li>\n<\/ul>\n\n\n\n
Recommending Products<\/h3>\n\n\n\n
Step 1: Calculate Similarity<\/h4>\n\n\n\n
When a user views a product, API generates embeddings for its title and description. <\/p>\n\n\n\n
Query the vector database using these embeddings to find similar products based on cosine similarity or other distance metrics.<\/p>\n\n\n\n
Step 2: Generate Results<\/h4>\n\n\n\n
The vector database retrieves products with the most similar embeddings, providing recommendations.<\/p>\n\n\n\n
Even if the recommended products are in different languages, the multilingual model ensures they share the same contextual meaning.<\/p>\n\n\n\n
Benefits of a Multilingual Recommender System<\/h3>\n\n\n\n
- \n
- Seamless User Experience:<\/strong> Users see recommendations in their preferred language.<\/li>\n\n\n\n
- Global Reach:<\/strong> Cater to diverse markets without language limitations.<\/li>\n\n\n\n
- Increased Sales:<\/strong> Personalized suggestions drive higher conversions.<\/li>\n\n\n\n
- Efficient Scaling:<\/strong> Multilingual embeddings reduce the need for separate models per language.<\/li>\n<\/ul>\n\n\n\n
You can also check out the Magento 2 Recommendation System<\/a>. It works with a multilingual embeddings model and supports dynamic models.<\/p>\n\n\n\n
<\/figure>\n\n\n\nThis is a pair of wireless headphones with the product description in English.<\/p>\n\n\n\n
The recommended products are available in different languages: the first in French, the second in Spanish, and the last in Italian.<\/p>\n\n\n\n
Our recommendation system suggests products even in different languages.<\/p>\n\n\n\n
Conclusion<\/h3>\n\n\n\n
A multilingual recommender system is vital for any e-commerce business targeting global markets.<\/p>\n\n\n\n
By leveraging multilingual embeddings, vector databases, and similarity metrics, you can create an inclusive and effective recommendation engine that resonates with users worldwide.<\/p>\n\n\n\n
- Global Reach:<\/strong> Cater to diverse markets without language limitations.<\/li>\n\n\n\n
- Pinecone<\/strong><\/li>\n\n\n\n
- ChromaDB<\/strong><\/li>\n\n\n\n