Find best matching item name

Hi @Jessica_Shin - possible solution in this post:

Not based on the number of words, but the “closeness” of one string to another.

This is the domain of full-text search with fuzzy logic, a capability that could be made possible as a formula if Airtable would finally allow us to create custom formulas from script blocks. Lacking this, it is possible to build a fuzzy search feature on top of Airtable bases.

Hi,
how do you expect to use it?
I have a small script doing something close (it checks 2 addresses from different sources) and put Y / N based on some ‘likeness’ into third field, not so precise as “Levenstein distance”, just words match, even without case align, but if you want to use it on the whole field, it would be even easier, than receive user input and generate text output.
I just tried to add UI part, learning deconstruction syntax, and I like it.

instead of (example from API doc):

const ordersTable = config.ordersTable;
const priceField = config.priceField;
const openOrdersView = config.openOrdersView;

using
const {ordersTable,priceField,openOrdersView}=config;

and here is the script

const config = input.config({title: 'Find best match',items:[
    input.config.table('table', {label: 'Select table'}),
    input.config.field('myfield', {label: 'Select field',parentTable: 'table'}),
    input.config.text('mytext',{label:'input text'})
    ]})
const {table,myfield:{name:field},mytext:text}=config;
const txtarr=[...new Set(text.split(' '))]
const reSet=x=>new Set(x.split(' '));
const sets=r=>[...txtarr,...reSet((r.getCellValueAsString(field)||''))]
const likeness=arr=>(arr.length-(reSet(arr.join(' '))).size)/arr.length
const compare=rec=>likeness(sets(rec))
const query=await table.selectRecordsAsync({fields:[field]});
const index=arr=>arr.indexOf(Math.max(...arr));
const val=query.records[index(query.records.map(compare))].getCellValueAsString(field);
output.text(val)