Envoyer du texte vers IRIS SQL avec LangChain

Une expérience sur la façon d'utiliser le framework LangChain, IRIS Vector Search et les LLM pour générer du SQL compatible IRIS à partir des invites utilisateur.

Cet article est basé sur ce cahier. Vous pouvez l'exécuter avec un environnement prêt à l'emploi avec cette application dans OpenExchange.

Installation

Tout d'abord, nous devons installer les bibliothèques nécessaires :

!pip install --upgrade --quiet langchain langchain-openai langchain-iris pandas

Ensuite, nous importons les modules requis et configurons l'environnement :

import os
import datetime
import hashlib
from copy import deepcopy
from sqlalchemy import create_engine
import getpass
import pandas as pd
from langchain_core.prompts import PromptTemplate, ChatPromptTemplate
from langchain_core.example_selectors import SemanticSimilarityExampleSelector
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain.docstore.document import Document
from langchain_community.document_loaders import DataFrameLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain_core.output_parsers import StrOutputParser
from langchain.globals import set_llm_cache
from langchain.cache import SQLiteCache
from langchain_iris import IRISVector

Nous utiliserons SQLiteCache pour mettre en cache les appels LLM :

# Cache for LLM calls
set_llm_cache(SQLiteCache(database_path=".langchain.db"))

Définissez les paramètres de connexion à la base de données IRIS :

# IRIS database connection parameters
os.environ["ISC_LOCAL_SQL_HOSTNAME"] = "localhost"
os.environ["ISC_LOCAL_SQL_PORT"] = "1972"
os.environ["ISC_LOCAL_SQL_NAMESPACE"] = "IRISAPP"
os.environ["ISC_LOCAL_SQL_USER"] = "_system"
os.environ["ISC_LOCAL_SQL_PWD"] = "SYS"

Si la clé API OpenAI n'est pas déjà définie dans l'environnement, invitez l'utilisateur à la saisir :

if not "OPENAI_API_KEY" in os.environ:
    os.environ["OPENAI_API_KEY"] = getpass.getpass()

Créez la chaîne de connexion pour la base de données IRIS :

# IRIS database connection string
args = {
    'hostname': os.getenv("ISC_LOCAL_SQL_HOSTNAME"), 
    'port': os.getenv("ISC_LOCAL_SQL_PORT"), 
    'namespace': os.getenv("ISC_LOCAL_SQL_NAMESPACE"), 
    'username': os.getenv("ISC_LOCAL_SQL_USER"), 
    'password': os.getenv("ISC_LOCAL_SQL_PWD")
}
iris_conn_str = f"iris://{args['username']}:{args['password']}@{args['hostname']}:{args['port']}/{args['namespace']}"

Établir la connexion à la base de données IRIS :

# Connection to IRIS database
engine = create_engine(iris_conn_str)
cnx = engine.connect().connection

Préparez un dictionnaire pour contenir les informations contextuelles pour l'invite système :

# Dict for context information for system prompt
context = {}
context["top_k"] = 3

Création d'invite

Pour transformer les entrées utilisateur en requêtes SQL compatibles avec la base de données IRIS, nous devons créer une invite efficace pour le modèle de langage. Nous commençons par une invite initiale qui fournit des instructions de base pour générer des requêtes SQL. Ce modèle est dérivé des invites par défaut de LangChain pour MSSQL et personnalisé pour la base de données IRIS.

# Basic prompt template with IRIS database SQL instructions
iris_sql_template = """
You are an InterSystems IRIS expert. Given an input question, first create a syntactically correct InterSystems IRIS query to run and return the answer to the input question.
Unless the user specifies in the question a specific number of examples to obtain, query for at most {top_k} results using the TOP clause as per InterSystems IRIS. You can order the results to return the most informative data in the database.
Never query for all columns from a table. You must query only the columns that are needed to answer the question. Wrap each column name in single quotes ('') to denote them as delimited identifiers.
Pay attention to use only the column names you can see in the tables below. Be careful to not query for columns that do not exist. Also, pay attention to which column is in which table.
Pay attention to use CAST(CURRENT_DATE as date) function to get the current date, if the question involves "today".
Use double quotes to delimit columns identifiers.
Return just plain SQL; don't apply any kind of formatting.
"""

Cette invite de base configure le modèle de langage (LLM) pour fonctionner comme un expert SQL avec des conseils spécifiques pour la base de données IRIS. Ensuite, nous fournissons une invite auxiliaire contenant des informations sur le schéma de la base de données pour éviter les hallucinations.

# SQL template extension for including tables context information
tables_prompt_template = """
Only use the following tables:
{table_info}
"""

Pour améliorer la précision des réponses du LLM, nous utilisons une technique appelée invite à quelques tirs. Il s'agit de présenter quelques exemples au LLM.

# SQL template extension for including few shots
prompt_sql_few_shots_template = """
Below are a number of examples of questions and their corresponding SQL queries.

{examples_value}
"""

Nous définissons le modèle pour quelques exemples :

# Few shots prompt template
example_prompt_template = "User input: {input}\nSQL query: {query}"
example_prompt = PromptTemplate.from_template(example_prompt_template)

Nous construisons l'invite utilisateur à l'aide du modèle en quelques plans :

# User prompt template
user_prompt = "\n" + example_prompt.invoke({"input": "{input}", "query": ""}).to_string()

Enfin, nous composons toutes les invites pour créer la finale :

# Complete prompt template
prompt = (
    ChatPromptTemplate.from_messages([("system", iris_sql_template)])
    + ChatPromptTemplate.from_messages([("system", tables_prompt_template)])
    + ChatPromptTemplate.from_messages([("system", prompt_sql_few_shots_template)])
    + ChatPromptTemplate.from_messages([("human", user_prompt)])
)
prompt

Cette invite attend les variables examples_value, input, table_info et top_k.

Voici comment l'invite est structurée :

ChatPromptTemplate(
    input_variables=['examples_value', 'input', 'table_info', 'top_k'], 
    messages=[
        SystemMessagePromptTemplate(
            prompt=PromptTemplate(
                input_variables=['top_k'], 
                template=iris_sql_template
            )
        ), 
        SystemMessagePromptTemplate(
            prompt=PromptTemplate(
                input_variables=['table_info'], 
                template=tables_prompt_template
            )
        ), 
        SystemMessagePromptTemplate(
            prompt=PromptTemplate(
                input_variables=['examples_value'], 
                template=prompt_sql_few_shots_template
            )
        ), 
        HumanMessagePromptTemplate(
            prompt=PromptTemplate(
                input_variables=['input'], 
                template=user_prompt
            )
        )
    ]
)

Pour visualiser comment l'invite sera envoyée au LLM, nous pouvons utiliser des valeurs d'espace réservé pour les variables requises :

prompt_value = prompt.invoke({
    "top_k": "<top_k>",
    "table_info": "<table_info>",
    "examples_value": "<examples_value>",
    "input": "<input>"
})
print(prompt_value.to_string())

System: 
You are an InterSystems IRIS expert. Given an input question, first create a syntactically correct InterSystems IRIS query to run and return the answer to the input question.
Unless the user specifies in the question a specific number of examples to obtain, query for at most <top_k> results using the TOP clause as per InterSystems IRIS. You can order the results to return the most informative data in the database.
Never query for all columns from a table. You must query only the columns that are needed to answer the question. Wrap each column name in single quotes ('') to denote them as delimited identifiers.
Pay attention to use only the column names you can see in the tables below. Be careful to not query for columns that do not exist. Also, pay attention to which column is in which table.
Pay attention to use CAST(CURRENT_DATE as date) function to get the current date, if the question involves "today".
Use double quotes to delimit columns identifiers.
Return just plain SQL; don't apply any kind of formatting.

System: 
Only use the following tables:
<table_info>

System: 
Below are a number of examples of questions and their corresponding SQL queries.

<examples_value>

Human: 
User input: <input>
SQL query: 

Maintenant, nous sommes prêts à envoyer cette invite au LLM en fournissant les variables nécessaires. Passons à l'étape suivante lorsque vous serez prêt.

Fournir des informations sur les tableaux

Pour créer des requêtes SQL précises, nous devons fournir au modèle de langage (LLM) des informations détaillées sur les tables de la base de données. Sans ces informations, le LLM pourrait générer des requêtes qui semblent plausibles mais incorrectes en raison d'hallucinations. Par conséquent, notre première étape consiste à créer une fonction qui récupère les définitions de table de la base de données IRIS.

Fonction pour récupérer les définitions de table

La fonction suivante interroge INFORMATION_SCHEMA pour obtenir les définitions de table pour un schéma spécifié. Si une table spécifique est fournie, il récupère la définition de cette table ; sinon, il récupère les définitions de toutes les tables du schéma.

def get_table_definitions_array(cnx, schema, table=None):
    cursor = cnx.cursor()

    # Base query to get columns information
    query = """
    SELECT TABLE_SCHEMA, TABLE_NAME, COLUMN_NAME, DATA_TYPE, IS_NULLABLE, COLUMN_DEFAULT, PRIMARY_KEY, null EXTRA
    FROM INFORMATION_SCHEMA.COLUMNS
    WHERE TABLE_SCHEMA = %s
    """

    # Parameters for the query
    params = [schema]

    # Adding optional filters
    if table:
        query += " AND TABLE_NAME = %s"
        params.append(table)

    # Execute the query
    cursor.execute(query, params)

    # Fetch the results
    rows = cursor.fetchall()

    # Process the results to generate the table definition(s)
    table_definitions = {}
    for row in rows:
        table_schema, table_name, column_name, column_type, is_nullable, column_default, column_key, extra = row
        if table_name not in table_definitions:
            table_definitions[table_name] = []
        table_definitions[table_name].append({
            "column_name": column_name,
            "column_type": column_type,
            "is_nullable": is_nullable,
            "column_default": column_default,
            "column_key": column_key,
            "extra": extra
        })

    primary_keys = {}

    # Build the output string
    result = []
    for table_name, columns in table_definitions.items():
        table_def = f"CREATE TABLE {schema}.{table_name} (\n"
        column_definitions = []
        for column in columns:
            column_def = f"  {column['column_name']} {column['column_type']}"
            if column['is_nullable'] == "NO":
                column_def += " NOT NULL"
            if column['column_default'] is not None:
                column_def += f" DEFAULT {column['column_default']}"
            if column['extra']:
                column_def += f" {column['extra']}"
            column_definitions.append(column_def)
        if table_name in primary_keys:
            pk_def = f"  PRIMARY KEY ({', '.join(primary_keys[table_name])})"
            column_definitions.append(pk_def)
        table_def += ",\n".join(column_definitions)
        table_def += "\n);"
        result.append(table_def)

    return result

Récupérer les définitions de table pour un schéma

Pour cet exemple, nous utilisons le schéma Aviation, disponible ici.

# Retrieve table definitions for the Aviation schema
tables = get_table_definitions_array(cnx, "Aviation")
print(tables)

Cette fonction renvoie les instructions CREATE TABLE pour toutes les tables du schéma Aviation :

[
    'CREATE TABLE Aviation.Aircraft (\n  Event bigint NOT NULL,\n  ID varchar NOT NULL,\n  AccidentExplosion varchar,\n  AccidentFire varchar,\n  AirFrameHours varchar,\n  AirFrameHoursSince varchar,\n  AirFrameHoursSinceLastInspection varchar,\n  AircraftCategory varchar,\n  AircraftCertMaxGrossWeight integer,\n  AircraftHomeBuilt varchar,\n  AircraftKey integer NOT NULL,\n  AircraftManufacturer varchar,\n  AircraftModel varchar,\n  AircraftRegistrationClass varchar,\n  AircraftSerialNo varchar,\n  AircraftSeries varchar,\n  Damage varchar,\n  DepartureAirportId varchar,\n  DepartureCity varchar,\n  DepartureCountry varchar,\n  DepartureSameAsEvent varchar,\n  DepartureState varchar,\n  DepartureTime integer,\n  DepartureTimeZone varchar,\n  DestinationAirportId varchar,\n  DestinationCity varchar,\n  DestinationCountry varchar,\n  DestinationSameAsLocal varchar,\n  DestinationState varchar,\n  EngineCount integer,\n  EvacuationOccurred varchar,\n  EventId varchar NOT NULL,\n  FlightMedical varchar,\n  FlightMedicalType varchar,\n  FlightPhase integer,\n  FlightPlan varchar,\n  FlightPlanActivated varchar,\n  FlightSiteSeeing varchar,\n  FlightType varchar,\n  GearType varchar,\n  LastInspectionDate timestamp,\n  LastInspectionType varchar,\n  Missing varchar,\n  OperationDomestic varchar,\n  OperationScheduled varchar,\n  OperationType varchar,\n  OperatorCertificate varchar,\n  OperatorCertificateNum varchar,\n  OperatorCode varchar,\n  OperatorCountry varchar,\n  OperatorIndividual varchar,\n  OperatorName varchar,\n  OperatorState varchar,\n  Owner varchar,\n  OwnerCertified varchar,\n  OwnerCountry varchar,\n  OwnerState varchar,\n  RegistrationNumber varchar,\n  ReportedToICAO varchar,\n  SeatsCabinCrew integer,\n  SeatsFlightCrew integer,\n  SeatsPassengers integer,\n  SeatsTotal integer,\n  SecondPilot varchar,\n  childsub bigint NOT NULL DEFAULT $i(^Aviation.EventC("Aircraft"))\n);',
    'CREATE TABLE Aviation.Crew (\n  Aircraft varchar NOT NULL,\n  ID varchar NOT NULL,\n  Age integer,\n  AircraftKey integer NOT NULL,\n  Category varchar,\n  CrewNumber integer NOT NULL,\n  EventId varchar NOT NULL,\n  Injury varchar,\n  MedicalCertification varchar,\n  MedicalCertificationDate timestamp,\n  MedicalCertificationValid varchar,\n  Seat varchar,\n  SeatbeltUsed varchar,\n  Sex varchar,\n  ShoulderHarnessUsed varchar,\n  ToxicologyTestPerformed varchar,\n  childsub bigint NOT NULL DEFAULT $i(^Aviation.AircraftC("Crew"))\n);',
    'CREATE TABLE Aviation.Event (\n  ID bigint NOT NULL DEFAULT $i(^Aviation.EventD),\n  AirportDirection integer,\n  AirportDistance varchar,\n  AirportElevation integer,\n  AirportLocation varchar,\n  AirportName varchar,\n  Altimeter varchar,\n  EventDate timestamp,\n  EventId varchar NOT NULL,\n  EventTime integer,\n  FAADistrictOffice varchar,\n  InjuriesGroundFatal integer,\n  InjuriesGroundMinor integer,\n  InjuriesGroundSerious integer,\n  InjuriesHighest varchar,\n  InjuriesTotal integer,\n  InjuriesTotalFatal integer,\n  InjuriesTotalMinor integer,\n  InjuriesTotalNone integer,\n  InjuriesTotalSerious integer,\n  InvestigatingAgency varchar,\n  LightConditions varchar,\n  LocationCity varchar,\n  LocationCoordsLatitude double,\n  LocationCoordsLongitude double,\n  LocationCountry varchar,\n  LocationSiteZipCode varchar,\n  LocationState varchar,\n  MidAir varchar,\n  NTSBId varchar,\n  NarrativeCause varchar,\n  NarrativeFull varchar,\n  NarrativeSummary varchar,\n  OnGroundCollision varchar,\n  SkyConditionCeiling varchar,\n  SkyConditionCeilingHeight integer,\n  SkyConditionNonCeiling varchar,\n  SkyConditionNonCeilingHeight integer,\n  TimeZone varchar,\n  Type varchar,\n  Visibility varchar,\n  WeatherAirTemperature integer,\n  WeatherPrecipitation varchar,\n  WindDirection integer,\n  WindDirectionIndicator varchar,\n  WindGust integer,\n  WindGustIndicator varchar,\n  WindVelocity integer,\n  WindVelocityIndicator varchar\n);'
]

Avec ces définitions de tableaux, nous pouvons passer à l'étape suivante, qui consiste à les intégrer dans notre invite pour le LLM. Cela garantit que le LLM dispose d'informations précises et complètes sur le schéma de base de données lors de la génération de requêtes SQL.

Sélection des tableaux les plus pertinents

Lorsque vous travaillez avec des bases de données, en particulier les plus volumineuses, l'envoi du langage de définition de données (DDL) pour toutes les tables dans une invite peut s'avérer peu pratique. Bien que cette approche puisse fonctionner pour de petites bases de données, les bases de données du monde réel contiennent souvent des centaines ou des milliers de tables, ce qui rend inefficace leur traitement global.

De plus, il est peu probable qu'un modèle de langage ait besoin de connaître chaque table de la base de données pour générer efficacement des requêtes SQL. Pour relever ce défi, nous pouvons exploiter les capacités de recherche sémantique pour sélectionner uniquement les tables les plus pertinentes en fonction de la requête de l'utilisateur.

Approche

Nous y parvenons en utilisant la recherche sémantique avec IRIS Vector Search. Notez que cette méthode est plus efficace si vos identifiants d'éléments SQL (tels que les tables, les champs et les clés) ont des noms significatifs. Si vos identifiants sont des codes arbitraires, pensez plutôt à utiliser un dictionnaire de données.

Mesures

1. Récupérer les informations du tableau

Tout d'abord, extrayez les définitions de table dans un DataFrame pandas :

# Retrieve table definitions into a pandas DataFrame
table_def = get_table_definitions_array(cnx=cnx, schema='Aviation')
table_df = pd.DataFrame(data=table_def, columns=["col_def"])
table_df["id"] = table_df.index + 1
table_df

The DataFrame (table_df) will look something like this:

	col_def	id
0	CREATE TABLE Aviation.Aircraft (\n Event bigi...	1
1	CREATE TABLE Aviation.Crew (\n Aircraft varch...	2
2	CREATE TABLE Aviation.Event (\n ID bigint NOT...	3

1. Split Definitions into Documents

Next, split the table definitions into Langchain Documents. This step is crucial for handling large chunks of text and extracting text embeddings:

loader = DataFrameLoader(table_df, page_content_column="col_def")
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=400, chunk_overlap=20, separator="\n")
tables_docs = text_splitter.split_documents(documents)
tables_docs

The resulting tables_docs list contains split documents with metadata, like so:

[Document(metadata={'id': 1}, page_content='CREATE TABLE Aviation.Aircraft (\n  Event bigint NOT NULL,\n  ID varchar NOT NULL,\n  ...'),
 Document(metadata={'id': 2}, page_content='CREATE TABLE Aviation.Crew (\n  Aircraft varchar NOT NULL,\n  ID varchar NOT NULL,\n  ...'),
 Document(metadata={'id': 3}, page_content='CREATE TABLE Aviation.Event (\n  ID bigint NOT NULL DEFAULT $i(^Aviation.EventD),\n  ...')]

1. Extract Embeddings and Store in IRIS

Now, use the IRISVector class from langchain-iris to extract embedding vectors and store them:

tables_vector_store = IRISVector.from_documents(
    embedding=OpenAIEmbeddings(), 
    documents=tables_docs,
    connection_string=iris_conn_str,
    collection_name="sql_tables",
    pre_delete_collection=True
)

Note: The pre_delete_collection flag is set to True for demonstration purposes to ensure a fresh collection in each test run. In a production environment, this flag should generally be set to False.

1. Find Relevant Documents

With the table embeddings stored, you can now query for relevant tables based on user input:

input_query = "List the first 2 manufacturers"
relevant_tables_docs = tables_vector_store.similarity_search(input_query, k=3)
relevant_tables_docs

For example, querying for manufacturers might return:

[Document(metadata={'id': 1}, page_content='GearType varchar,\n  LastInspectionDate timestamp,\n  ...'),
 Document(metadata={'id': 1}, page_content='AircraftModel varchar,\n  AircraftRegistrationClass varchar,\n  ...'),
 Document(metadata={'id': 3}, page_content='LocationSiteZipCode varchar,\n  LocationState varchar,\n  ...')]

From the metadata, you can see that only table ID 1 (Aviation.Aircraft) is relevant, which aligns with the query.

1. Handling Edge Cases

While this approach is generally effective, it may not always be perfect. For instance, querying for crash sites might also return less relevant tables:

input_query = "List the top 10 most crash sites"
relevant_tables_docs = tables_vector_store.similarity_search(input_query, k=3)
relevant_tables_docs

Results might include:

[Document(metadata={'id': 3}, page_content='LocationSiteZipCode varchar,\n  LocationState varchar,\n  ...'),
 Document(metadata={'id': 3}, page_content='InjuriesGroundSerious integer,\n  InjuriesHighest varchar,\n  ...'),
 Document(metadata={'id': 1}, page_content='CREATE TABLE Aviation.Aircraft (\n  Event bigint NOT NULL,\n  ID varchar NOT NULL,\n  ...')]

Despite retrieving the correct Aviation.Event table twice, the Aviation.Aircraft table may also appear, which could be improved with additional filtering or thresholding. This is beyond the scope of this example and will be left for future implementations.

1. Define a Function to Retrieve Relevant Tables

To automate this process, define a function to filter and return the relevant tables based on user input:

def get_relevant_tables(user_input, tables_vector_store, table_df):
    relevant_tables_docs = tables_vector_store.similarity_search(user_input)
    relevant_tables_docs_indices = [x.metadata["id"] for x in relevant_tables_docs]
    indices = table_df["id"].isin(relevant_tables_docs_indices)
    relevant_tables_array = [x for x in table_df[indices]["col_def"]]
    return relevant_tables_array

This function will help in efficiently retrieving only the relevant tables to send to the LLM, reducing the prompt length and improving overall query performance.

Selecting the Most Relevant Examples (Few-Shot Prompting)

When working with language models (LLMs), providing them with relevant examples helps ensure accurate and contextually appropriate responses. These examples, referred to as "few-shot" examples, guide the LLM in understanding the structure and context of the queries it should handle.

In our case, we need to populate the examples_value variable with a diverse set of SQL queries that cover a broad spectrum of IRIS SQL syntax and the tables available in the database. This helps prevent the LLM from generating incorrect or irrelevant queries.

Defining Example Queries

Below is a list of example queries designed to illustrate various SQL operations:

examples = [
    {"input": "List all aircrafts.", "query": "SELECT * FROM Aviation.Aircraft"},
    {"input": "Find all incidents for the aircraft with ID 'N12345'.", "query": "SELECT * FROM Aviation.Event WHERE EventId IN (SELECT EventId FROM Aviation.Aircraft WHERE ID = 'N12345')"},
    {"input": "List all incidents in the 'Commercial' operation type.", "query": "SELECT * FROM Aviation.Event WHERE EventId IN (SELECT EventId FROM Aviation.Aircraft WHERE OperationType = 'Commercial')"},
    {"input": "Find the total number of incidents.", "query": "SELECT COUNT(*) FROM Aviation.Event"},
    {"input": "List all incidents that occurred in 'Canada'.", "query": "SELECT * FROM Aviation.Event WHERE LocationCountry = 'Canada'"},
    {"input": "How many incidents are associated with the aircraft with AircraftKey 5?", "query": "SELECT COUNT(*) FROM Aviation.Aircraft WHERE AircraftKey = 5"},
    {"input": "Find the total number of distinct aircrafts involved in incidents.", "query": "SELECT COUNT(DISTINCT AircraftKey) FROM Aviation.Aircraft"},
    {"input": "List all incidents that occurred after 5 PM.", "query": "SELECT * FROM Aviation.Event WHERE EventTime > 1700"},
    {"input": "Who are the top 5 operators by the number of incidents?", "query": "SELECT TOP 5 OperatorName, COUNT(*) AS IncidentCount FROM Aviation.Aircraft GROUP BY OperatorName ORDER BY IncidentCount DESC"},
    {"input": "Which incidents occurred in the year 2020?", "query": "SELECT * FROM Aviation.Event WHERE YEAR(EventDate) = '2020'"},
    {"input": "What was the month with most events in the year 2020?", "query": "SELECT TOP 1 MONTH(EventDate) EventMonth, COUNT(*) EventCount FROM Aviation.Event WHERE YEAR(EventDate) = '2020' GROUP BY MONTH(EventDate) ORDER BY EventCount DESC"},
    {"input": "How many crew members were involved in incidents?", "query": "SELECT COUNT(*) FROM Aviation.Crew"},
    {"input": "List all incidents with detailed aircraft information for incidents that occurred in the year 2012.", "query": "SELECT e.EventId, e.EventDate, a.AircraftManufacturer, a.AircraftModel, a.AircraftCategory FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE Year(e.EventDate) = 2012"},
    {"input": "Find all incidents where there were more than 5 injuries and include the aircraft manufacturer and model.", "query": "SELECT e.EventId, e.InjuriesTotal, a.AircraftManufacturer, a.AircraftModel FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE e.InjuriesTotal > 5"},
    {"input": "List all crew members involved in incidents with serious injuries, along with the incident date and location.", "query": "SELECT c.CrewNumber AS 'Crew Number', c.Age, c.Sex AS Gender, e.EventDate AS 'Event Date', e.LocationCity AS 'Location City', e.LocationState AS 'Location State' FROM Aviation.Crew c JOIN Aviation.Event e ON c.EventId = e.EventId WHERE c.Injury = 'Serious'"}
]

Selecting Relevant Examples

Given the ever-expanding list of examples, it’s impractical to provide the LLM with all of them. Instead, we use IRIS Vector Search along with the SemanticSimilarityExampleSelector class to identify the most relevant examples based on user prompts.

Define the Example Selector:

example_selector = SemanticSimilarityExampleSelector.from_examples(
    examples,
    OpenAIEmbeddings(),
    IRISVector,
    k=5,
    input_keys=["input"],
    connection_string=iris_conn_str,
    collection_name="sql_samples",
    pre_delete_collection=True
)

Note: The pre_delete_collection flag is used here for demonstration purposes to ensure a fresh collection in each test run. In a production environment, this flag should be set to False to avoid unnecessary deletions.

Query the Selector:

To find the most relevant examples for a given input, use the selector as follows:

input_query = "Find all events in 2010 informing the Event Id and date, location city and state, aircraft manufacturer and model."
relevant_examples = example_selector.select_examples({"input": input_query})

The results might look like this:

[{'input': 'List all incidents with detailed aircraft information for incidents that occurred in the year 2012.', 'query': 'SELECT e.EventId, e.EventDate, a.AircraftManufacturer, a.AircraftModel, a.AircraftCategory FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE Year(e.EventDate) = 2012'},
 {'input': "Find all incidents for the aircraft with ID 'N12345'.", 'query': "SELECT * FROM Aviation.Event WHERE EventId IN (SELECT EventId FROM Aviation.Aircraft WHERE ID = 'N12345')"},
 {'input': 'Find all incidents where there were more than 5 injuries and include the aircraft manufacturer and model.', 'query': 'SELECT e.EventId, e.InjuriesTotal, a.AircraftManufacturer, a.AircraftModel FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE e.InjuriesTotal > 5'},
 {'input': 'List all aircrafts.', 'query': 'SELECT * FROM Aviation.Aircraft'},
 {'input': 'Find the total number of distinct aircrafts involved in incidents.', 'query': 'SELECT COUNT(DISTINCT AircraftKey) FROM Aviation.Aircraft'}]

If you specifically need examples related to quantities, you can query the selector accordingly:

input_query = "What is the number of incidents involving Boeing aircraft."
quantity_examples = example_selector.select_examples({"input": input_query})

The output may be:

[{'input': 'How many incidents are associated with the aircraft with AircraftKey 5?', 'query': 'SELECT COUNT(*) FROM Aviation.Aircraft WHERE AircraftKey = 5'},
 {'input': 'Find the total number of distinct aircrafts involved in incidents.', 'query': 'SELECT COUNT(DISTINCT AircraftKey) FROM Aviation.Aircraft'},
 {'input': 'How many crew members were involved in incidents?', 'query': 'SELECT COUNT(*) FROM Aviation.Crew'},
 {'input': 'Find all incidents where there were more than 5 injuries and include the aircraft manufacturer and model.', 'query': 'SELECT e.EventId, e.InjuriesTotal, a.AircraftManufacturer, a.AircraftModel FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE e.InjuriesTotal > 5'},
 {'input': 'List all incidents with detailed aircraft information for incidents that occurred in the year 2012.', 'query': 'SELECT e.EventId, e.EventDate, a.AircraftManufacturer, a.AircraftModel, a.AircraftCategory FROM Aviation.Event e JOIN Aviation.Aircraft a ON e.EventId = a.EventId WHERE Year(e.EventDate) = 2012'}]

This output includes examples that specifically address counting and quantities.

Future Considerations

While the SemanticSimilarityExampleSelector is powerful, it’s important to note that not all selected examples may be perfect. Future improvements may involve adding filters or thresholds to exclude less relevant results, ensuring that only the most appropriate examples are provided to the LLM.

Accuracy Test

To assess the performance of the prompt and SQL query generation, we need to set up and run a series of tests. The goal is to evaluate how well the LLM generates SQL queries based on user inputs, with and without the use of example-based few shots.

Function to Generate SQL Queries

We start by defining a function that uses the LLM to generate SQL queries based on the provided context, prompt, user input, and other parameters:

def get_sql_from_text(context, prompt, user_input, use_few_shots, tables_vector_store, table_df, example_selector=None, example_prompt=None):
    relevant_tables = get_relevant_tables(user_input, tables_vector_store, table_df)
    context["table_info"] = "\n\n".join(relevant_tables)

    examples = example_selector.select_examples({"input": user_input}) if example_selector else []
    context["examples_value"] = "\n\n".join([
        example_prompt.invoke(x).to_string() for x in examples
    ])

    model = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
    output_parser = StrOutputParser()
    chain_model = prompt | model | output_parser

    response = chain_model.invoke({
        "top_k": context["top_k"],
        "table_info": context["table_info"],
        "examples_value": context["examples_value"],
        "input": user_input
    })
    return response

Execute the Prompt

Test the prompt with and without examples:

# Prompt execution **with** few shots
input = "Find all events in 2010 informing the Event Id and date, location city and state, aircraft manufacturer and model."
response_with_few_shots = get_sql_from_text(
    context, 
    prompt, 
    user_input=input, 
    use_few_shots=True, 
    tables_vector_store=tables_vector_store, 
    table_df=table_df,
    example_selector=example_selector, 
    example_prompt=example_prompt,
)
print(response_with_few_shots)

SELECT e.EventId, e.EventDate, e.LocationCity, e.LocationState, a.AircraftManufacturer, a.AircraftModel
FROM Aviation.Event e
JOIN Aviation.Aircraft a ON e.EventId = a.EventId
WHERE Year(e.EventDate) = 2010

# Prompt execution **without** few shots
input = "Find all events in 2010 informing the Event Id and date, location city and state, aircraft manufacturer and model."
response_with_no_few_shots = get_sql_from_text(
    context, 
    prompt, 
    user_input=input, 
    use_few_shots=False, 
    tables_vector_store=tables_vector_store, 
    table_df=table_df,
)
print(response_with_no_few_shots)

SELECT TOP 3 "EventId", "EventDate", "LocationCity", "LocationState", "AircraftManufacturer", "AircraftModel"
FROM Aviation.Event e
JOIN Aviation.Aircraft a ON e.ID = a.Event
WHERE e.EventDate >= '2010-01-01' AND e.EventDate < '2011-01-01'

Utility Functions for Testing

To test the generated SQL queries, we define some utility functions:

def execute_sql_query(cnx, query):
    try:
        cursor = cnx.cursor()
        cursor.execute(query)
        rows = cursor.fetchall()
        return rows
    except:
        print('Error running query:')
        print(query)
        print('-'*80)
    return None

def sql_result_equals(cnx, query, expected):
    rows = execute_sql_query(cnx, query)
    result = [set(row._asdict().values()) for row in rows or []]
    if result != expected and rows is not None:
        print('Result not as expected for query:')
        print(query)
        print('-'*80)
    return result == expected

# SQL test for prompt **with** few shots
print("SQL is OK" if not execute_sql_query(cnx, response_with_few_shots) is None else "SQL is not OK")

    SQL is OK

# SQL test for prompt **without** few shots
print("SQL is OK" if not execute_sql_query(cnx, response_with_no_few_shots) is None else "SQL is not OK")

    error on running query: 
    SELECT TOP 3 "EventId", "EventDate", "LocationCity", "LocationState", "AircraftManufacturer", "AircraftModel"
    FROM Aviation.Event e
    JOIN Aviation.Aircraft a ON e.ID = a.Event
    WHERE e.EventDate >= '2010-01-01' AND e.EventDate < '2011-01-01'
    --------------------------------------------------------------------------------
    SQL is not OK

Define and Execute Tests

Define a set of test cases and run them:

tests = [{
    "input": "What were the top 3 years with the most recorded events?",
    "expected": [{128, 2003}, {122, 2007}, {117, 2005}]
},{
    "input": "How many incidents involving Boeing aircraft.",
    "expected": [{5}]
},{
    "input": "How many incidents that resulted in fatalities.",
    "expected": [{237}]
},{
    "input": "List event Id and date and, crew number, age and gender for incidents that occurred in 2013.",
    "expected": [{1, datetime.datetime(2013, 3, 4, 11, 6), '20130305X71252', 59, 'M'},
                 {1, datetime.datetime(2013, 1, 1, 15, 0), '20130101X94035', 32, 'M'},
                 {2, datetime.datetime(2013, 1, 1, 15, 0), '20130101X94035', 35, 'M'},
                 {1, datetime.datetime(2013, 1, 12, 15, 0), '20130113X42535', 25, 'M'},
                 {2, datetime.datetime(2013, 1, 12, 15, 0), '20130113X42535', 34, 'M'},
                 {1, datetime.datetime(2013, 2, 1, 15, 0), '20130203X53401', 29, 'M'},
                 {1, datetime.datetime(2013, 2, 15, 15, 0), '20130218X70747', 27, 'M'},
                 {1, datetime.datetime(2013, 3, 2, 15, 0), '20130303X21011', 49, 'M'},
                 {1, datetime.datetime(2013, 3, 23, 13, 52), '20130326X85150', 'M', None}]
},{
    "input": "Find the total number of incidents that occurred in the United States.",
    "expected": [{1178}]
},{
    "input": "List all incidents latitude and longitude coordinates with more than 5 injuries that occurred in 2010.",
    "expected": [{-78.76833333333333, 43.25277777777778}]
},{
    "input": "Find all incidents in 2010 informing the Event Id and date, location city and state, aircraft manufacturer and model.",
    "expected": [
        {datetime.datetime(2010, 5, 20, 13, 43), '20100520X60222', 'CIRRUS DESIGN CORP', 'Farmingdale', 'New York', 'SR22'},
        {datetime.datetime(2010, 4, 11, 15, 0), '20100411X73253', 'CZECH AIRCRAFT WORKS SPOL SRO', 'Millbrook', 'New York', 'SPORTCRUISER'},
        {'108', datetime.datetime(2010, 1, 9, 12, 55), '20100111X41106', 'Bayport', 'New York', 'STINSON'},
        {datetime.datetime(2010, 8, 1, 14, 20), '20100801X85218', 'A185F', 'CESSNA', 'New York', 'Newfane'}
    ]
}]

Accuracy Evaluation

Run the tests and calculate the accuracy:

def execute_tests(cnx, context, prompt, use_few_shots, tables_vector_store, table_df, example_selector, example_prompt):
    tests_generated_sql = [(x, get_sql_from_text(
            context, 
            prompt, 
            user_input=x['input'], 
            use_few_shots=use_few_shots, 
            tables_vector_store=tables_vector_store, 
            table_df=table_df,
            example_selector=example_selector if use_few_shots else None, 
            example_prompt=example_prompt if use_few_shots else None,
        )) for x in deepcopy(tests)]

    tests_sql_executions = [(x[0], sql_result_equals(cnx, x[1], x[0]['expected'])) 
                            for x in tests_generated_sql]

    accuracy = sum(1 for i in tests_sql_executions if i[1] == True) / len(tests_sql_executions)
    print(f'Accuracy: {accuracy}')
    print('-'*80)

Results

# Accuracy tests for prompts executed **without** few shots
use_few_shots = False
execute_tests(
    cnx,
    context, 
    prompt, 
    use_few_shots, 
    tables_vector_store, 
    table_df, 
    example_selector, 
    example_prompt
)

    error on running query: 
    SELECT "EventDate", COUNT("EventId") as "TotalEvents"
    FROM Aviation.Event
    GROUP BY "EventDate"
    ORDER BY "TotalEvents" DESC
    TOP 3;
    --------------------------------------------------------------------------------
    error on running query: 
    SELECT "EventId", "EventDate", "C"."CrewNumber", "C"."Age", "C"."Sex"
    FROM "Aviation.Event" AS "E"
    JOIN "Aviation.Crew" AS "C" ON "E"."ID" = "C"."EventId"
    WHERE "E"."EventDate" >= '2013-01-01' AND "E"."EventDate" < '2014-01-01'
    --------------------------------------------------------------------------------
    result not expected for query: 
    SELECT TOP 3 "e"."EventId", "e"."EventDate", "e"."LocationCity", "e"."LocationState", "a"."AircraftManufacturer", "a"."AircraftModel"
    FROM "Aviation"."Event" AS "e"
    JOIN "Aviation"."Aircraft" AS "a" ON "e"."ID" = "a"."Event"
    WHERE "e"."EventDate" >= '2010-01-01' AND "e"."EventDate" < '2011-01-01'
    --------------------------------------------------------------------------------
    accuracy: 0.5714285714285714
    --------------------------------------------------------------------------------

# Accuracy tests for prompts executed **with** few shots
use_few_shots = True
execute_tests(
    cnx,
    context, 
    prompt, 
    use_few_shots, 
    tables_vector_store, 
    table_df, 
    example_selector, 
    example_prompt
)

    error on running query: 
    SELECT e.EventId, e.EventDate, e.LocationCity, e.LocationState, a.AircraftManufacturer, a.AircraftModel
    FROM Aviation.Event e
    JOIN Aviation.Aircraft a ON e.EventId = a.EventId
    WHERE Year(e.EventDate) = 2010 TOP 3
    --------------------------------------------------------------------------------
    accuracy: 0.8571428571428571
    --------------------------------------------------------------------------------

Conclusion

The accuracy of SQL queries generated with examples (few shots) is approximately 49% higher compared to those generated without examples (85% vs. 57%).

References

• https://python.langchain.com/v0.1/docs/expression_language/get_started/
• https://python.langchain.com/v0.1/docs/use_cases/sql/prompting/
• https://python.langchain.com/v0.1/docs/modules/model_io/prompts/composition/

Ce qui précède est le contenu détaillé de. pour plus d'informations, suivez d'autres articles connexes sur le site Web de PHP en chinois!