create synthetic data

martygubar · martygubar · commit 091aa8242b79 · 2024-11-22T17:21:11.000-05:00
diff --git a/sql/README.md b/sql/README.md
@@ -18,15 +18,15 @@ Try out these scripts to learn how to get started using Autonomous Database. Sim
 |----|---|
 |[credential-create.sql](credential-create.sql)|Autonomous Database credentials contain the secret keys used to connect to services - like Azure OpenAI. This script creates those credentials. It's called by multiple scripts listed below |
 |[data-create-sample-schema.sql](data-create-sample-schema.sql)|Create a sample user and install sample data|
-|[data-create-synthetic-data.sql](data-create-synthetic-data.sql)|Use AI to generate sample data sets|
+|[data-create-synthetic-data.sql](data-create-synthetic-data.sql)|Use AI to generate sample data sets. You will create tables and then populate them based on rules that you specify.|
 |[data-import-from-datalake.sql](data-create-synthetic-data.sql)|Import sample data from Azure Data Lake. Sample data was uploaded using the [`create-all-resources.sh`](../azure-cli/create-all-resources.sh)and [`create-data-lake-storage.sh`](../azure-cli/create-data-lake-storage.sh) scripts. You can run [`show-data-lake-storage-info.sh`](../azure-cli/show-data-lake-storage-info.sh) to get connection information to the storage container.|
 |[data-export-to-datalake.sql](data-export-to-datalake.sql)|Export data from a table to your data lake storage|
 |[select-ai-admin-enable.sql](select-ai-admin-enable.sql)|Autonomous Database is secure by default. In order to access external services, you will need to enable connectivity. This script enables connectivity to your Azure OpenAI resource.|
 |[select-ai-create-profile.sql](select-ai-create-profile.sql)|A Select AI profile is used to connect to your AI provider. It includes information about your provider plus tables and view that you want to be targets for natural language queries|
 |[select-ai-nl2sql.sql](select-ai-nl2sql.sql)|Use natural language to query your data|
 |[select-ai-sql-function.sql](select-ai-sql-function.sql)|Use Select AI SQL functions to apply AI to your data. These examples summarize a support chat and make product recommendations based on info in your database|
-|[select-ai-rag.sql](select-ai-rag.sql)|Select AI makes it easy to create AI vector pipelines and then ask questions using AI and your organization's knowledge base|
-|[json-duality.sql](json-duality.sql)|Autonomous Database allows you to with JSON in many different ways. JSON Collections stored in the database, and JSON Duality Views are a couple of the examples here.|
+|[select-ai-rag.sql](select-ai-rag.sql)|Select AI makes it easy to create AI vector management pipelines and then ask questions using AI and your organization's knowledge base|
+|[json-duality.sql](json-duality.sql)|Autonomous Database allows you to work with JSON in many different ways. JSON Collections stored your documents in tables. No changes are made to those documents. JSON Duality Views present relational data as JSON - allowing you to query and update that data using the JSON APIs (like MongoDB API) or SQL. Check out both approaches.|
 
 ### Configuration file
 Many of the SQL scripts rely on information found in your [config.sql](config.sql) file. Update the config file prior to running any of the scripts. 
diff --git a/sql/data-create-synthetic-data.sql b/sql/data-create-synthetic-data.sql
@@ -1,2 +1,111 @@
 -- Copyright (c) 2024 Oracle and/or its affiliates.
 -- Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
+
+/* PREREQUISITES
+    Install the sample schema using script
+        @data-create-sample-schema.sql
+*/
+
+-- config.sql contains the endpoints, resource groups and other settings required to connect to your Azure OpenAI deployment
+@config.sql
+
+-- Create your credential and Select AI Profile
+@select-ai-create-profile.sql
+
+drop table if exists activity cascade constraints;
+drop table if exists customers cascade constraints;
+drop table if exists award_winners cascade constraints;
+
+-- Create a table containing award winning movies
+CREATE TABLE award_winners AS
+SELECT *
+FROM movies
+WHERE instr(awards, 'Academy') > 0;
+
+SELECT title, awards
+FROM AWARD_WINNERS;
+
+
+/*
+  Add PK constraints to existing tables. This will ensure that data generation for related tables will create valid values.
+  For example, ACTIVITY data will be generated - and valid MOVIE_ID's will be generated for that table b/c of this primary key
+*/
+ALTER TABLE award_winners ADD CONSTRAINT constraint_name PRIMARY KEY (movie_id) ENABLE;
+
+/* Create and generate customers */
+CREATE TABLE customers (	
+    cust_id NUMBER, 
+    last_name VARCHAR2(100), 
+    first_name VARCHAR2(100), 
+    email VARCHAR2(200), 
+    street_address VARCHAR2(4000), 
+    postal_code VARCHAR2(10), 
+    city VARCHAR2(100), 
+    state_province VARCHAR2(100), 
+    country VARCHAR2(400), 
+    yrs_customer NUMBER, 
+    age_range varchar2(4000), 
+    gender VARCHAR2(4000), 
+    household_size NUMBER, 
+  
+    CONSTRAINT pk_customers PRIMARY KEY (cust_id)
+  )
+;	
+
+-- These comments will tell the AI model how to generate data for a column
+COMMENT ON COLUMN customers.gender IS 'The value for gender should be either: Male, Female, Non-binary';
+COMMENT ON COLUMN customers.YRS_CUSTOMER IS 'The value for YRS_CUSTOMER should be: number between 1 and 8';
+COMMENT ON COLUMN customers.age_range IS 'The value for age_range should be either: Silent Generation, Baby Boomer, GenX, Millenials, GenZ';
+
+CREATE TABLE activity
+   (
+    day_id TIMESTAMP (6), 
+	movie_id NUMBER, 
+	cust_id NUMBER, 
+	app VARCHAR2(100), 
+	device VARCHAR2(100), 
+	list_price NUMBER(5,2), 
+	discount_percent NUMBER(5,2), 
+	amount NUMBER(5,2),  
+  CONSTRAINT fk_activity_cust_id FOREIGN KEY (cust_id)
+    REFERENCES customers (cust_id) ENABLE,
+  CONSTRAINT fk_activity_movie_id FOREIGN KEY (movie_id)
+    REFERENCES award_winners (movie_id) ENABLE
+   );     
+
+-- These comments will tell the AI model how to generate data for a column
+COMMENT ON COLUMN activity.day_id IS 'Date between 01-JAN-2023 and 30-SEP-2024';
+COMMENT ON COLUMN activity.device IS 'Possible values: laptop, roku, appletv';
+COMMENT ON COLUMN activity.list_price IS 'Possible values: 0,1.99,3.99';
+COMMENT ON COLUMN activity.discount_percent IS 'Fraction between 0 and 1. If list_price is 0, then 0';
+COMMENT ON COLUMN activity.amount IS 'Value equals list_price * 1-discount';
+
+--
+-- Generate data based on the rules in the comments. You can add additional rules to the generate function too.
+-- The data will be generated based on the rules in the column contents
+-- And, data integrity rules are applied for the primary->foreign key relationships
+BEGIN
+    DBMS_CLOUD_AI.GENERATE_SYNTHETIC_DATA(
+        profile_name => '&AZURE_OPENAI_PROFILE_NAME',  
+        object_list => '[                          
+                          {"owner": "&USER_NAME", "name": "customers","record_count":20},
+                          {"owner": "&USER_NAME", "name": "activity","record_count":5500}
+                        ]',
+		params => '{
+                "comments":true,
+                "priority":"HIGH"
+                }'
+    );
+END;
+/
+
+-- Query the new data sets!
+select ai what are total sales and discounts by age range;
+select ai list 10 customers include all attributes;
+
+
+
+
+
+select * from "MOVIESTREAM"."AWARD_WINNERS";
+
diff --git a/sql/property-graph.sql b/sql/property-graph.sql
@@ -0,0 +1,223 @@
+CREATE TABLE MOVIES (
+    MOVIE_ID        NUMBER,
+    TITLE           VARCHAR2(400),
+    GENRES          JSON,
+    SUMMARY         VARCHAR2(16000)
+);
+
+CREATE TABLE MOVIES_CUSTOMER (
+    CUST_ID        NUMBER,
+    FIRSTNAME      VARCHAR(200),
+    LASTNAME       VARCHAR(200)
+);
+
+CREATE TABLE WATCHED_MOVIE (
+    DAY_ID            TIMESTAMP(6),      
+    MOVIE_ID          NUMBER,
+    PROMO_CUST_ID     NUMBER
+);
+
+CREATE TABLE WATCHED_WITH (
+    ID                NUMBER,
+    WP_ID             NUMBER,
+    WATCHER           NUMBER,
+    WATCHED_WITH      NUMBER,
+    DATE_WATCHED      TIMESTAMP(6),
+    MOVIE_ID          NUMBER,
+    WATCH_PARTY_NAME  VARCHAR(200)
+);
+
+INSERT INTO MOVIES (MOVIE_ID, TITLE, GENRES, SUMMARY) VALUES
+    (1, 'Inception', '{"Action": "Sci-Fi"}', 'A thief who steals corporate secrets through the use of dream-sharing technology is given the inverse task of planting an idea into the mind of a C.E.O.'),
+    (2, 'The Matrix', '{"Action": "Sci-Fi"}', 'A computer hacker learns from mysterious rebels about the true nature of his reality and his role in the war against its controllers.'),
+    (3, 'The Godfather', '{"Drama": "Crime"}', 'The aging patriarch of an organized crime dynasty transfers control of his clandestine empire to his reluctant son.'),
+    (4, 'Titanic', '{"Romance": "Drama"}', 'A seventeen-year-old aristocrat falls in love with a kind but poor artist aboard the luxurious, ill-fated R.M.S. Titanic.'),
+    (5, 'Toy Story', '{"Animation": "Adventure"}', 'A cowboy doll is profoundly threatened and jealous when a new spaceman figure supplants him as top toy in a boy''s room.');
+
+INSERT INTO MOVIES_CUSTOMER (CUST_ID, FIRSTNAME, LASTNAME) VALUES
+    (101, 'John', 'Doe'),
+    (102, 'Jane', 'Smith'),
+    (103, 'Sam', 'Wilson'),
+    (104, 'Emily', 'Clark'),
+    (105, 'Michael', 'Johnson');
+
+INSERT INTO WATCHED_MOVIE (DAY_ID, MOVIE_ID, PROMO_CUST_ID) VALUES
+    (TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 101),
+    (TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 102),
+    (TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 103),
+    (TO_TIMESTAMP('2024-10-31 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 2, 101),
+    (TO_TIMESTAMP('2024-09-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 101),
+    (TO_TIMESTAMP('2024-10-31 09:15:23.654321', 'YYYY-MM-DD HH24:MI:SS.FF'), 2, 102),
+    (TO_TIMESTAMP('2024-11-01 16:45:12.987654', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 103),
+    (TO_TIMESTAMP('2024-11-02 18:22:43.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 4, 104),
+    (TO_TIMESTAMP('2024-11-03 20:01:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF'), 5, 105);
+
+INSERT INTO WATCHED_WITH (ID, WP_ID, WATCHER, WATCHED_WITH, DATE_WATCHED, MOVIE_ID, WATCH_PARTY_NAME) VALUES 
+    (1, 101, 101, 102, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 'New Year Party'),
+    (2, 101, 101, 103, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 'New Year Party'),
+    (3, 101, 102, 103, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 1, 'New Year Party'),
+    (4, 102, 101, 103, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 'Doe Watch Party'),
+    (5, 102, 101, 103, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 'Doe Watch Party'),
+    (6, 102, 104, 103, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 'Doe Watch Party'),
+    (7, 102, 101, 104, TO_TIMESTAMP('2024-10-30 12:34:56.123456', 'YYYY-MM-DD HH24:MI:SS.FF'), 3, 'Doe Watch Party');
+
+
+BEGIN
+DBMS_CLOUD.CREATE_CREDENTIAL (
+    credential_name => 'GRAPH_CREDENTIAL',
+    user_ocid       => 'ocid1.user.oc1..aaaaaaaan4vamdkpzj4mcayzihhbdbty27bsomnlpmshu5hcy3w6ceqfwieq',
+    tenancy_ocid    => 'ocid1.tenancy.oc1..aaaaaaaaj4ccqe763dizkrcdbs5x7ufvmmojd24mb6utvkymyo4xwxyv3gfa',
+    private_key     => 'MIIEvAIBADANBgkqhkiG9w0BAQEFAASCBKYwggSiAgEAAoIBAQCDoD9ZzA5BMbtx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',
+fingerprint     => '5d:cf:7e:55:16:20:e1:60:f6:31:57:4d:ab:f7:35:98');
+END;
+/
+
+BEGIN   
+     -- drops the profile if it already exists
+     DBMS_CLOUD_AI.drop_profile(profile_name => 'genai', force => true);
+
+     -- Create an AI profile that uses the default LLAMA model on OCI
+     dbms_cloud_ai.create_profile(
+         profile_name => 'genai',
+         attributes =>       
+             '{"provider": "oci",
+             "credential_name": "GRAPH_CREDENTIAL",
+             "comments":"true",            
+             "oci_compartment_id": "ocid1.compartment.oc1..aaaaaaaajlby3soifhkjobwner46lpjf5zvwugnzrmdfymtcdu6r2lvtti7a",
+             "region":"us-chicago-1",
+             "object_list": [
+                 {"owner": "ADMIN", "name": "MOVIES"},
+                 {"owner": "ADMIN", "name": "MOVIES_CUSTOMER"},
+                 {"owner": "ADMIN", "name": "WATCHED_MOVIE"},
+                 {"owner": "ADMIN", "name": "WATCHED_WITH"}
+             ]
+             }'
+         );
+ END;
+
+
+
+CREATE PROPERTY GRAPH CUSTOMER_WATCHED_MOVIES
+        VERTEX TABLES (
+            MOVIES_CUSTOMER AS CUSTOMER
+                KEY(CUST_ID),
+            MOVIES AS MOVIE
+                KEY(MOVIE_ID)
+        )
+        EDGE TABLES(
+            WATCHED_MOVIE AS WATCHED
+                KEY(DAY_ID, MOVIE_ID, PROMO_CUST_ID)
+                SOURCE KEY (PROMO_CUST_ID) REFERENCES CUSTOMER(CUST_ID)
+                DESTINATION KEY (MOVIE_ID) REFERENCES MOVIE(MOVIE_ID),
+            WATCHED_WITH
+                KEY(ID)
+                SOURCE KEY (WATCHER) REFERENCES CUSTOMER(CUST_ID)
+                DESTINATION KEY (WATCHED_WITH) REFERENCES CUSTOMER(CUST_ID)
+        );
+
+-- Get the Movie title and summary for each movie a customer has watched
+-- We can send this to an LLM as a Graph RAG solution to answer questions like 
+--- "What is the genre of {movie} based on this summary"
+SELECT DISTINCT MOVIE_TITLE, MOVIE_SUMMARY
+    FROM GRAPH_TABLE( CUSTOMER_WATCHED_MOVIES
+        MATCH (c1 IS CUSTOMER)-[e1 IS WATCHED]->(m IS MOVIE)
+        COLUMNS (m.title as MOVIE_TITLE, m.summary as MOVIE_SUMMARY) 
+    );
+
+WITH prompt_document AS (
+    SELECT
+        JSON_OBJECT ('TASK' VALUE 'What is the genre of this movie based on this summary',
+        MOVIE_TITLE,
+        MOVIE_SUMMARY ) AS prompt_details
+        FROM GRAPH_TABLE(CUSTOMER_WATCHED_MOVIES
+        MATCH (c1 IS CUSTOMER)-[e1 IS WATCHED]->(m IS MOVIE)
+        COLUMNS (m.title as MOVIE_TITLE, m.summary as MOVIE_SUMMARY) 
+    )
+)
+SELECT 
+    DBMS_LOB.SUBSTR(DBMS_CLOUD_AI.GENERATE(
+        PROMPT => prompt_details,
+        PROFILE_NAME => 'genai',
+        ACTION       => 'chat'
+    ), 4000, 1) AS Answer
+FROM prompt_document;
+
+-- Find customers who are connected through 2/3 watch party connections
+SELECT DISTINCT *
+    FROM GRAPH_TABLE( CUSTOMER_WATCHED_MOVIES
+        MATCH (c1 IS CUSTOMER)-[e1 IS WATCHED_WITH]-{2,3}(c2 IS CUSTOMER)
+        COLUMNS (c1.CUST_ID as c1_cust_id, c1.FIRSTNAME as c1_fist_name, c2.cust_id as c2_cust_id, c2.FIRSTNAME as c2_fist_name) 
+    );
+
+-- Find customers who have watched the same movie and are connected in 2/3 hops, excluding results where customer 1 and 2 are the same 
+-- This can be sent to an LLM as part of a Graph RAG solution to answer questions like
+-- "Based on this dataset of movies customers who are connected through friends of friends have watched, containing customer IDs, names, movie titles, genres and summaries. How would you describe the movie watching preferences of these users?"
+SELECT DISTINCT *
+    FROM GRAPH_TABLE(CUSTOMER_WATCHED_MOVIES
+        MATCH (c1 IS CUSTOMER)-[e1 IS WATCHED_WITH]-{2,3}(c2 IS CUSTOMER) -[e2 is WATCHED]-> (m is MOVIE),
+              (c1) -[e3 is WATCHED]-> (m is MOVIE)
+        WHERE c1.cust_id <> c2.cust_id
+        COLUMNS (c1.CUST_ID as c1_cust_id, c1.FIRSTNAME as c1_fist_name, c2.cust_id as c2_cust_id, c2.FIRSTNAME as c2_fist_name, m.title as movie_title, m.genres as movie_genre, m.summary as movie_summary) 
+    );
+
+
+-- Create the JSON object with task and result set
+WITH prompt_document AS (
+    SELECT
+        JSON_OBJECT(
+            'TASK' VALUE 'Based on this dataset of movies customers who are connected through friends of friends have watched, containing customer IDs, names, movie titles, genres and summaries. How would you describe the movie watching preferences of these users?',
+            'PROMPT_DETAILS' VALUE JSON_ARRAYAGG(
+                JSON_OBJECT(
+                    'c1_cust_id' VALUE c1_cust_id,
+                    'c1_fist_name' VALUE c1_fist_name,
+                    'c2_cust_id' VALUE c2_cust_id,
+                    'movie_title' VALUE movie_title,
+                    'movie_genre' VALUE movie_genre,
+                    'movie_summary' VALUE movie_summary
+                ) RETURNING CLOB
+            ) RETURNING CLOB
+        ) AS prompt_details
+    FROM GRAPH_TABLE(
+        CUSTOMER_WATCHED_MOVIES
+        MATCH (c1 IS CUSTOMER)-[e1 IS WATCHED_WITH]-{2,3}(c2 IS CUSTOMER) -[e2 is WATCHED]-> (m is MOVIE),
+              (c1) -[e3 is WATCHED]-> (m is MOVIE)
+        WHERE c1.cust_id <> c2.cust_id
+        COLUMNS (c1.CUST_ID as c1_cust_id, c1.FIRSTNAME as c1_fist_name, c2.cust_id as c2_cust_id, c2.FIRSTNAME as c2_fist_name, m.title as movie_title, m.genres as movie_genre, m.summary as movie_summary) 
+    )
+)
+SELECT 
+    DBMS_LOB.SUBSTR(
+        DBMS_CLOUD_AI.GENERATE(
+            PROMPT => prompt_details,
+            PROFILE_NAME => 'genai',
+            ACTION       => 'chat'
+        ),
+        4000,
+        1
+    ) AS Answer
+FROM prompt_document;
