Cette page décrit la structure recommandée pour rédiger des prompts efficaces pour vos agents de données qui utilisent l'API Conversational Analytics et se connectent aux données BigQuery. Ces prompts sont des contextes créés que vous définissez sous forme de chaînes à l'aide du paramètre system_instruction.
Exemples de composants de clés pour les instructions système
Les sections suivantes présentent des exemples de composants de clés pour les instructions système dans BigQuery. Voici certaines de ces clés :
tablesfieldsmeasuresgolden_queriesgolden_action_plansrelationshipsglossariesadditional_descriptions
Pour obtenir une description de ces composants de clés, consultez la page de documentation Guider le comportement de l'agent avec un contexte créé.
Décrire vos données avec tables
Le bloc de code YAML suivant montre la structure de base de la clé tables pour la table bigquery-public-data.thelook_ecommerce.orders :
- tables:
- table:
- name: bigquery-public-data.thelook_ecommerce.orders
- description: Data for customer orders in The Look fictitious e-commerce store.
- synonyms:
- sales
- orders_data
- tags:
- ecommerce
- transaction
Décrire les champs couramment utilisés avec fields
L'exemple de code YAML suivant décrit des champs de clés tels que order_id, status, created_at, num_of_items et earnings pour la table orders :
- tables:
- table:
- name: bigquery-public-data.thelook_ecommerce.orders
- fields:
- field:
- name: order_id
- description: The unique identifier for each customer order.
- field:
- name: user_id
- description: The unique identifier for each customer.
- field:
- name: status
- description: The current status of the order.
- sample_values:
- complete
- shipped
- returned
- field:
- name: created_at
- description: The timestamp when the order was created.
- field:
- name: num_of_items
- description: The total number of items in the order.
- aggregations:
- sum
- avg
- field:
- name: earnings
- description: The sales amount for the order.
- aggregations:
- sum
- avg
Définir des métriques métier avec measures
Par exemple, vous pouvez définir une mesure profit en tant que soustraction des coûts aux revenus comme suit :
- tables:
- table:
- name: bigquery-public-data.thelook_ecommerce.orders
- measures:
- measure:
- name: profit
- description: Raw profit (earnings minus cost).
- exp: earnings - cost
- synonyms: gains
Améliorer la précision avec golden_queries
Par exemple, vous pouvez définir des requêtes de référence pour les analyses courantes des données de la table orders comme suit :
- tables:
- table:
- golden_queries:
- golden_query:
- natural_language_query: How many orders are there?
- sql_query: SELECT COUNT(*) FROM sqlgen-testing.thelook_ecommerce.orders
- golden_query:
- natural_language_query: How many orders were shipped?
- sql_query: >-
SELECT COUNT(*) FROM sqlgen-testing.thelook_ecommerce.orders
WHERE status = 'shipped'
Décrire des tâches en plusieurs étapes avec golden_action_plans
Par exemple, vous pouvez définir un plan d'action pour afficher la répartition des commandes par tranche d'âge en incluant des informations sur les étapes liées à la requête SQL et à la visualisation :
- tables:
- table:
- golden_action_plans:
- golden_action_plan:
- natural_language_query: Show me the number of orders broken down by age group.
- action_plan:
- step: >-
Run a SQL query that joins the table
sqlgen-testing.thelook_ecommerce.orders and
sqlgen-testing.thelook_ecommerce.users to get a
breakdown of order count by age group.
- step: >-
Create a vertical bar plot using the retrieved data,
with one bar per age group.
Définir des jointures de tables avec relationships
Par exemple, vous pouvez définir une relation orders_to_user entre la table bigquery-public-data.thelook_ecommerce.orders et la table bigquery-public-data.thelook_ecommerce.users comme suit :
- relationships:
- relationship:
- name: orders_to_user
- description: >-
Connects customer order data to user information with the user_id and id fields to allow an aggregated view of sales by customer demographics.
- relationship_type: many-to-one
- join_type: left
- left_table: bigquery-public-data.thelook_ecommerce.orders
- right_table: bigquery-public-data.thelook_ecommerce.users
- relationship_columns:
- left_column: user_id
- right_column: id
Expliquer les termes métier avec glossaries
Par exemple, vous pouvez définir des termes tels que les états d'activité courants et "OMPF" en fonction de votre contexte métier spécifique comme suit :
- glossaries:
- glossary:
- term: complete
- description: Represents an order status where the order has been completed.
- synonyms: 'finish, done, fulfilled'
- glossary:
- term: shipped
- description: Represents an order status where the order has been shipped to the customer.
- glossary:
- term: returned
- description: Represents an order status where the customer has returned the order.
- glossary:
- term: OMPF
- description: Order Management and Product Fulfillment
Inclure des instructions supplémentaires avec additional_descriptions
Par exemple, vous pouvez utiliser la clé additional_descriptions pour fournir des informations sur votre organisation comme suit :
- additional_descriptions:
- text: All the sales data pertains to The Look, a fictitious ecommerce store.
- text: 'Orders can be of three categories: food, clothes, and electronics.'
Exemple : Instructions système dans BigQuery
L'exemple suivant montre des instructions système pour un agent devant se comporter comme un analyste des ventes :
- system_instruction: >-
You are an expert sales analyst for a fictitious ecommerce store. You will answer questions about sales, orders, and customer data. Your responses should be concise and data-driven.
- tables:
- table:
- name: bigquery-public-data.thelook_ecommerce.orders
- description: Data for orders in The Look, a fictitious ecommerce store.
- synonyms: sales
- tags: 'sale, order, sales_order'
- fields:
- field:
- name: order_id
- description: The unique identifier for each customer order.
- field:
- name: user_id
- description: The unique identifier for each customer.
- field:
- name: status
- description: The current status of the order.
- sample_values:
- complete
- shipped
- returned
- field:
- name: created_at
- description: >-
The date and time at which the order was created in timestamp
format.
- field:
- name: returned_at
- description: >-
The date and time at which the order was returned in timestamp
format.
- field:
- name: num_of_items
- description: The total number of items in the order.
- aggregations: 'sum, avg'
- field:
- name: earnings
- description: The sales revenue for the order.
- aggregations: 'sum, avg'
- field:
- name: cost
- description: The cost for the items in the order.
- aggregations: 'sum, avg'
- measures:
- measure:
- name: profit
- description: Raw profit (earnings minus cost).
- exp: earnings - cost
- synonyms: gains
- golden_queries:
- golden_query:
- natural_language_query: How many orders are there?
- sql_query: SELECT COUNT(*) FROM sqlgen-testing.thelook_ecommerce.orders
- golden_query:
- natural_language_query: How many orders were shipped?
- sql_query: >-
SELECT COUNT(*) FROM sqlgen-testing.thelook_ecommerce.orders
WHERE status = 'shipped'
- golden_action_plans:
- golden_action_plan:
- natural_language_query: Show me the number of orders broken down by age group.
- action_plan:
- step: >-
Run a SQL query that joins the table
sqlgen-testing.thelook_ecommerce.orders and
sqlgen-testing.thelook_ecommerce.users to get a
breakdown of order count by age group.
- step: >-
Create a vertical bar plot using the retrieved data,
with one bar per age group.
- table:
- name: bigquery-public-data.thelook_ecommerce.users
- description: Data for users in The Look, a fictitious ecommerce store.
- synonyms: customers
- tags: 'user, customer, buyer'
- fields:
- field:
- name: id
- description: The unique identifier for each user.
- field:
- name: first_name
- description: The first name of the user.
- tag: person
- sample_values: 'alex, izumi, nur'
- field:
- name: last_name
- description: The first name of the user.
- tag: person
- sample_values: 'warmer, stilles, smith'
- field:
- name: age_group
- description: The age demographic group of the user.
- sample_values:
- 18-24
- 25-34
- 35-49
- 50+
- field:
- name: email
- description: The email address of the user.
- tag: contact
- sample_values: '222larabrown@gmail.com, cloudysanfrancisco@gmail.com'
- golden_queries:
- golden_query:
- natural_language_query: How many unique customers are there?
- sql_query: >-
SELECT COUNT(DISTINCT id) FROM
bigquery-public-data.thelook_ecommerce.users
- golden_query:
- natural_language_query: How many users in the 25-34 age group have a cymbalgroup email address?
- sql_query: >-
SELECT COUNT(DISTINCT id) FROM
bigquery-public-data.thelook_ecommerce.users WHERE users.age_group =
'25-34' AND users.email LIKE '%@cymbalgroup.com';
- relationships:
- relationship:
- name: orders_to_user
- description: >-
Connects customer order data to user information with the user_id and id fields to allow an aggregated view of sales by customer demographics.
- relationship_type: many-to-one
- join_type: left
- left_table: bigquery-public-data.thelook_ecommerce.orders
- right_table: bigquery-public-data.thelook_ecommerce.users
- relationship_columns:
- left_column: user_id
- right_column: id
- glossaries:
- glossary:
- term: complete
- description: Represents an order status where the order has been completed.
- synonyms: 'finish, done, fulfilled'
- glossary:
- term: shipped
- description: Represents an order status where the order has been shipped to the customer.
- glossary:
- term: returned
- description: Represents an order status where the customer has returned the order.
- glossary:
- term: OMPF
- description: Order Management and Product Fulfillment
- additional_descriptions:
- text: All the sales data pertains to The Look, a fictitious ecommerce store.
- text: 'Orders can be of three categories: food, clothes, and electronics.'