Mises à jour et suppressions en temps réel
Apprenez des techniques pour gérer les données dynamiques, notamment les mises à jour et suppressions efficaces de vecteurs en temps réel.
Mises à jour et suppressions en temps réel est une leçon Vector Databases: Pinecone, Weaviate & pgvector gratuite sur CoddyKit. Ceci est la leçon 3 sur 4. Tu peux lire la leçon complète ci-dessous gratuitement — puis la pratiquer en direct dans le navigateur avec un éditeur de code intégré et un tuteur IA 24/7. Elle fait partie du parcours d'apprentissage Vector Databases: Pinecone, Weaviate & pgvector, et ta progression se synchronise sur le web et l'application CoddyKit. Le cours Vector Databases: Pinecone, Weaviate & pgvector comprend 4 leçons au total.
Certaines parties de cette leçon n'ont pas encore été traduites et s'affichent en anglais.
Handling Dynamic Vector Data
In real-world applications, data isn't static. It constantly changes! This means the vectors representing your data also need to be updated or removed from your vector database.
Imagine a product catalog where prices change, or user profiles where preferences evolve. Your vector database must reflect these changes to remain accurate.
Why Real-time Changes Matter
Real-time updates and deletions are crucial for maintaining the integrity and relevance of your applications:
- Accuracy: Ensuring your search results are always based on the freshest data.
- Relevance: Keeping recommendations and contextual information up-to-date for users.
- Data Governance: Complying with data retention policies or user requests to delete personal data (e.g., 'right to be forgotten').
Full Vector & Metadata Updates
To completely replace an existing vector's embedding values and its associated metadata in Pinecone, you use the upsert() method.
If a vector with the same ID already exists in your index, upsert() will overwrite its existing vector values and metadata with the new ones you provide. If no vector with that ID exists, it will be inserted as a new entry.
Example: Replacing a Vector
If a product's description changes significantly, its embedding might need a full replacement. We use upsert() with the existing ID to update both the vector and its metadata.
from pinecone import Pinecone, Index
import os
# For this example, we'll use a mock class instead of actual Pinecone connection
# In a real app, replace with:
# pc = Pinecone(api_key=os.environ.get("PINECONE_API_KEY"), environment=os.environ.get("PINECONE_ENVIRONMENT"))
# index = pc.Index("your-index-name")
class MockPineconeIndex:
def upsert(self, vectors):
print(f"Upserting vectors: {vectors}")
index = MockPineconeIndex()
# Assume 'product-123' already exists with an old vector and metadata
print("--- Before Update (Conceptual) ---")
print("Vector 'product-123' has old values and metadata.")
# New vector values and metadata for 'product-123'
new_vector_values = [0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2] # Example 8-dim vector
new_metadata = {"category": "electronics", "price": 129.99, "status": "updated"}
# Use upsert to fully replace the vector and metadata for 'product-123'
index.upsert(
vectors=[
{"id": "product-123", "values": new_vector_values, "metadata": new_metadata}
]
)
print("\n--- After Update (Conceptual) ---")
print("Vector 'product-123' now has new values and metadata, replacing old ones.")Partial Updates with `update()`
Sometimes you only need to change a vector's metadata, or perhaps just its vector values without touching metadata. Pinecone provides an update() method for these partial modifications.
Using update() is more efficient than upsert() when you only have partial changes, as you only send the data that has changed, rather than replacing the entire entry.
Example: Updating Metadata Only
To update only the metadata for an existing vector, you use the update() method. You provide the vector's ID and the set_metadata parameter with the new metadata. The existing vector values will remain untouched.
from pinecone import Pinecone, Index
import os
# Mock class for Pinecone's update method
class MockPineconeIndexUpdate:
def update(self, id, values=None, set_metadata=None):
print(f"Updating vector ID: {id}")
if values: print(f" New values provided (vector will be replaced): {values}")
if set_metadata: print(f" Setting metadata: {set_metadata}")
index_update = MockPineconeIndexUpdate() # Replace with pc.Index("your-index-name")
# Assume 'product-456' exists with vector and metadata like {'stock': 10, 'color': 'blue'}
print("--- Before Metadata Update (Conceptual) ---")
print("Vector 'product-456' has existing metadata.")
# Update only the stock count for 'product-456'
updated_metadata = {"stock": 5, "last_checked": "2023-10-27"}
index_update.update(id="product-456", set_metadata=updated_metadata)
print("\n--- After Metadata Update (Conceptual) ---")
print("Vector 'product-456' now has updated stock and last_checked metadata. Other metadata and vector values are unchanged.")Deleting Vectors in Pinecone
Removing vectors from your Pinecone index is a common operation. Pinecone offers flexible ways to delete vectors:
- By one or more specific ID(s).
- By matching a metadata filter.
- Deleting all vectors within an index or a specific namespace.
This functionality is vital for data hygiene, managing outdated information, and adhering to privacy regulations.
Example: Deleting a Single Vector
The most straightforward way to delete a vector is by its unique ID. You pass a list of one or more IDs to the delete() method.
from pinecone import Pinecone, Index
import os
# Mock class for Pinecone's delete method
class MockPineconeIndexDelete:
def delete(self, ids=None, delete_all=False, filter=None):
if ids: print(f"Deleting vectors with IDs: {ids}")
elif delete_all: print("Deleting all vectors.")
elif filter: print(f"Deleting vectors with filter: {filter}")
index_delete = MockPineconeIndexDelete() # Replace with pc.Index("your-index-name")
# Assume 'user-profile-789' exists in the index
print("--- Before Deletion (Conceptual) ---")
print("Vector 'user-profile-789' is present.")
# Delete a single vector by its ID
index_delete.delete(ids=["user-profile-789"])
print("\n--- After Deletion (Conceptual) ---")
print("Vector 'user-profile-789' has been removed from the index.")Deleting Many Vectors by Filter
While you can delete multiple vectors by providing a list of IDs, a powerful feature is deleting vectors based on a metadata filter.
This allows you to efficiently remove all vectors that match specific metadata criteria, such as all products from a certain category, or all inactive user profiles, without knowing their individual IDs.
Example: Deleting with Filters
Let's say we want to remove all vectors associated with products marked as 'out_of_stock'. We can use a filter object within the delete() method.
from pinecone import Pinecone, Index
import os
# Mock class for Pinecone's delete method
class MockPineconeIndexDelete:
def delete(self, ids=None, delete_all=False, filter=None):
if ids: print(f"Deleting vectors with IDs: {ids}")
elif delete_all: print("Deleting all vectors.")
elif filter: print(f"Deleting vectors with filter: {filter}")
index_delete = MockPineconeIndexDelete() # Replace with pc.Index("your-index-name")
# Assume several products exist, some with 'status': 'out_of_stock'
print("--- Before Filtered Deletion (Conceptual) ---")
print("Vectors for products with status 'out_of_stock' are present.")
# Delete all vectors where the 'status' metadata is 'out_of_stock'
index_delete.delete(filter={
"status": "out_of_stock"
})
print("\n--- After Filtered Deletion (Conceptual) ---")
print("All vectors for 'out_of_stock' products have been removed.")Check Your Understanding
Which of the following statements about updating and deleting vectors in Pinecone are TRUE?
Recap & Next Steps
Great job! You've learned how to keep your Pinecone index dynamic and up-to-date.
- We explored how
upsert()performs a full replacement of vectors and metadata. - You saw how the
update()method is used for efficient partial changes, especially for metadata. - Finally, we covered various ways to delete vectors, including by ID and using powerful metadata filters.
These operations are essential for managing evolving datasets in real-world AI applications.
Questions Fréquemment Posées
La leçon « Mises à jour et suppressions en temps réel » est-elle gratuite ?
Oui — le texte complet de « Mises à jour et suppressions en temps réel » est gratuit à lire ici sur le web. Pour la pratiquer de manière interactive (un éditeur de code intégré et un tuteur IA 24/7) et déverrouiller le reste du cours Vector Databases: Pinecone, Weaviate & pgvector, passe à CoddyKit PRO. Le cours Vector Databases: Pinecone, Weaviate & pgvector comprend 4 leçons au total.
Qu'est-ce que j'apprendrai dans « Mises à jour et suppressions en temps réel » ?
Apprenez des techniques pour gérer les données dynamiques, notamment les mises à jour et suppressions efficaces de vecteurs en temps réel. Tu pratiques Vector Databases: Pinecone, Weaviate & pgvector avec du code pratique que tu exécutes directement dans le navigateur, et un tuteur IA 24/7 répond à tes questions au fur et à mesure que tu avances dans la leçon.
Dois-je avoir de l'expérience pour commencer Vector Databases: Pinecone, Weaviate & pgvector ?
Aucune expérience préalable n'est requise. Vector Databases: Pinecone, Weaviate & pgvector sur CoddyKit est structuré pour les débutants jusqu'aux apprenants avancés, donc tu peux commencer ici ou depuis le début et avancer à ton rythme. Ceci est la leçon 3 sur 4.
Combien de temps prend la leçon « Mises à jour et suppressions en temps réel » ?
La plupart des leçons CoddyKit prennent environ 5–10 minutes. Chacune est courte et interactive, tu progresses régulièrement et tu repiques exactement où tu t'es arrêté sur le web et l'app.
Peux-tu écrire et exécuter du code dans cette leçon Vector Databases: Pinecone, Weaviate & pgvector ?
Oui. Chaque leçon Vector Databases: Pinecone, Weaviate & pgvector inclut un éditeur de code intégré, tu écris et exécutes du vrai code directement dans ton navigateur et tu reçois des retours IA instantanés — aucune configuration locale requise.
Toutes les leçons de ce cours
- Filtrer avec les métadonnées
- Gérer les espaces de noms
- Mises à jour et suppressions en temps réel
- Recherche hybride avec vecteurs creux et denses