Very interesting, but not what I expected when I heard the word “ChatGPT”. In fact I don’t even see evidence of an LLM at work here, just AI-generated SQL queries.
Q:How many repeating articles were posted in 2023?
I'm a robot running Auto-GPT on TiDB Cloud Serverless
Our goals:
Answer: How many repeating articles were posted in 2023?
Shutdown upon achieving goals
Thoughts: Let's list all tables and check if 'items' table is in the database
Reasoning: We have to find out if the 'items' table exists because it contains the information we need to answer the question 'How many repeating articles were posted in 2023?'
Plan:
Use the command 'tidb_query_executor' to list all tables in the database
Check if 'items' table is in the result
Criticism: None
Next action:
Command = tidb_query_executor
Query =
SHOW TABLES;
I'm waiting for your feedback...
Command tidb_query_executor returned:
Tables_in_hackernews
items
users
Thoughts: We've confirmed that the 'items' table exists in the database, so let's examine its schema to determine which columns we should use.
Reasoning: By examining the schema for the 'items' table, we'll be able to see which columns are relevant for answering the question 'How many repeating articles were posted in 2023?'
Plan:
Use the command 'tidb_query_executor' to list the schema for the 'items' table
Identify the columns needed to answer the question
Criticism: None
Next action:
Command = tidb_query_executor
Query =
SHOW CREATE TABLE `items`;
I'm waiting for your feedback...
Command tidb_query_executor returned:
Table Create Table
items CREATE TABLE items (
id bigint(20) NOT NULL COMMENT 'The item''s unique id.',
title varchar(198) DEFAULT NULL COMMENT 'The title of the story, poll or job. HTML.',
url varchar(6598) DEFAULT NULL COMMENT 'The URL of the story.',
text text DEFAULT NULL COMMENT 'The comment, story or poll text. HTML.',
type varchar(8) NOT NULL DEFAULT 'story' COMMENT 'The type of item. One of "job", "story", "comment", "poll", or "pollopt".',
by varchar(15) NOT NULL DEFAULT '' COMMENT 'The username of the item''s author.',
time int(11) NOT NULL COMMENT 'Creation date of the item, in Unix Time.',
score int(11) NOT NULL DEFAULT '0' COMMENT 'The story''s score, or the votes for a pollopt.\n',
parent bigint(20) DEFAULT NULL COMMENT 'The comment''s parent: either another comment or the relevant story.',
descendants int(11) DEFAULT '0' COMMENT 'In the case of stories or polls, the total comment count.',
deleted tinyint(1) NOT NULL DEFAULT '0' COMMENT 'true if the item is deleted.',
dead tinyint(1) NOT NULL DEFAULT '0' COMMENT 'true if the item is dead.',
last_fetch_at datetime NOT NULL DEFAULT '1970-01-01 00:00:01',
kids json DEFAULT NULL COMMENT 'The ids of the item''s comments, in ranked display order.',
parts json DEFAULT NULL COMMENT 'A list of related pollopts, in display order.',
poll bigint(20) DEFAULT NULL COMMENT 'The pollopt''s associated poll.',
PRIMARY KEY (id) /T![clustered_index] CLUSTERED /,
KEY index_items_type_time_id (type,time,id),
KEY index_items_time (time),
KEY index_items_by (by),
KEY index_items_score (score),
KEY index_items_type_deleted_dead_time (type,deleted,dead,time),
KEY index_items_descendants (descendants),
KEY index_items_deleted_dead_time_score (deleted,dead,time,score),
KEY index_items_score1 (score),
KEY index_items_score2 (score)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin
Thoughts: Now we know which columns we need to query to answer 'How many repeating articles were posted in 2023?', so let's generate SQL to query the 'items' table
Reasoning: By selecting the 'title', 'time', and 'url' columns from the 'items' table, we can determine which articles were posted in 2023 and if they are repeating
Plan:
Use the SELECT command to query the 'items' table
Add the WHERE clause to filter the results to the articles posted in 2023
Use GROUP BY and COUNT to identify the repeating articles
Criticism: None
Next action:
Command = tidb_query_executor
Query =
SELECT title, time, url, COUNT() FROM items WHERE type = 'story' AND time BETWEEN UNIX_TIMESTAMP('2023-01-01 00:00:00') AND UNIX_TIMESTAMP('2023-12-31 23:59:59') GROUP BY title, url HAVING COUNT() > 1 ORDER BY COUNT(*) DESC;
I'm waiting for your feedback...
