What is this tool for?
In large information systems (banks, insurers, public bodies), the same piece of data should keep the same column name from one table to another. Before creating a table, the tool checks each column: it already exists (and in which tables), or it is new. For a new column, it looks for the closest known names: often a typo (AMOUTN), a different abbreviation (INT_RT for INTEREST_RATE) or words in another order.
Frequently asked questions
Are my schemas sent anywhere?
No. All processing happens in your browser; pasted tables are only kept on your computer for next time (use “Clear all” to remove them). You can use the tool on confidential schemas.
Which formats can I paste?
CREATE TABLE scripts (Oracle, SQL Server, PostgreSQL, MySQL, DB2…), a CSV export of INFORMATION_SCHEMA.COLUMNS or an equivalent view (ALL_TAB_COLUMNS in Oracle), a copy from Excel (table, column, type) or a plain list of columns, with a TABLE: NAME line before each table. You can also open a .zip of extracts: one CSV file per table, whose name gives the table name and whose first line gives the columns. Only this header line is read; the data in the files is neither read in full nor kept.
How are similar names found?
Names are first brought to a common form (upper case, snake_case and CamelCase treated alike, prefixes such as T_ ignored). The tool then compares letters (Damerau-Levenshtein distance: missing, extra or swapped letter), words after replacing abbreviations (editable dictionary), word order and shared words. Each suggestion shows a similarity percentage and the reason.
Can I add my company's vocabulary?
Yes: the abbreviation dictionary can be edited in the settings (“AMOUNT = AMT, AMNT”), exported to a text file and imported by a colleague. It only contains words, no information about your tables.
What does the rename script contain?
For each column where you chose a known name, an ALTER TABLE … RENAME COLUMN statement (or sp_rename for SQL Server). Review it before running it: the tool suggests, you decide.
Checking a new table in five minutes
- Export the existing columns. In Oracle:
SELECT TABLE_NAME, COLUMN_NAME, DATA_TYPE FROM ALL_TAB_COLUMNS WHERE OWNER = 'MY_SCHEMA'; in SQL Server, PostgreSQL or MySQL: the same query onINFORMATION_SCHEMA.COLUMNS. Copy the result (or the CSV file) into “Reference tables”.CREATE TABLEscripts work too, and so does a.zipof CSV extracts (one file per table): each file name gives the table and its header line the columns; data rows are not read. - Paste the new table's columns: its
CREATE TABLEscript or just the list of names. - Read the result: known columns are listed separately; for each missing column, the closest known names appear with their similarity percentage, the reason (typo, abbreviation, swapped words, shared words) and the tables where they are found.
- Decide: click the right suggestion, or choose “New column”.
- Export the report (CSV, opens in Excel) for review, and the rename script to fix the table.
Examples of matches
LOAN_AMOUTN→LOAN_AMOUNT: two letters swapped (typo).NOMINAL_INT_RATE→NOMINAL_INTEREST_RATE: same words, different abbreviations.CUSTOMER_CODEandCODE_CUSTOMER: same words in another order.ClosingDateandCLOSING_DATE: the same name written in CamelCase, recognised as existing.
Tuning the tool
- Minimum similarity: at 60% the tool suggests widely; raise it to 75–80% on very large schemas to keep only likely matches.
- Prefixes to ignore: add your naming standard's prefixes (
T,V,TB…) so they don't affect comparison. - Dictionary: add your company's vocabulary, then export it for colleagues. The richer it is, the more abbreviations are recognised.
- Types: if your exports include types, tick the option to rule out matching a date with an amount, and to spot an existing column declared with another type.
Guide written and checked by Erwan Hautot · updated 7 November 2026.