What's the best way to scan a column with long text and update a column in the same record with a set of multiple select 'tags' or options?

[Khushi_Suri]

I’d like to find a way to construct a column that scans the Title column and updates a column in the same record with multiple select options.

To further explain, I’m building a database that has newspaper/press headlines (integrated via RSS feed and zapier). Each record has a “title” that has the heading of the article. I’d like to find a way to automatically update a column in the same record with multiple select options based on what text is in the name column.

For example (within the same record) if the “Title” column of a record says “Three birds were seen flying from the eagle’s nest on Tuesday and are now migrating north to Montreal.” - the tagged fields column would read this and output (birds)(migration)(Canada).

Is this possible on airtable? Where would I start?

[Tim_Sullivan]

Take a look at this post by @Bill.French.

Polish your Data with Auto-Tagging Show and tell

The other day a friend of mine watched as I imported 7,000 support tickets from Jira into Airtable and then auto-tagged them in less than a minute. He was amazed; I was yawning. [image] I suppose it’s a pretty cool thing, but we often get good at something(s) and forget about how much time they save and possibly how much more useful the data becomes when we attend to the seemingly little things like consistent keyword tagging through automation. I use my Airdrop framework to do this - it’s a…

[kuovonne]

Bill’s script is a great place to start. However, it extracts keywords that actually appear in the text. Notice that you want keywords that do not appear in your text.

Khushi_Suri:

“Three birds were seen flying from the eagle’s nest on Tuesday and are now migrating north to Montreal.” - the tagged fields column would read this and output (birds)(migration)(Canada).

Bill’s method could extract “birds”. It would also extract “migrating”, not “migration” because they are different words. It also would not be able to infer “Canada” from “Montreal”. Plus you would get a lot of other keywords.

This is not a criticism of Bill’s script, which is pretty amazing. Rather, I’m trying to point out the difficulty of what you are trying to do. If you really need keywords that do not appear in your text, I suggest you find a third party service that specializes in keyword extraction, then have a script that hooks into their api. (If you find a service that you like but need help writing that script, I offer professional script writing services.)