How to use to_tsvector with no-dictionary (no stemming)?

Question

I am using to_tsvector, and it works fine... But a new demand, to preserve exact original words (raw text), need "bypass" as dictionary.

... something like to use to_tsvector('raw', myString) where myString is something like "AATT GAA", no meaning for any dictionary.

`to_tsvector('simple', myString)` – Daniel Vérité Mar 03 '16 at 17:32 — Daniel Vérité, Mar 03 '16 at 17:32

score 5 · Accepted Answer · answered Mar 03 '16 at 19:42

5

SELECT to_tsvector('simple','your string');

answered Mar 03 '16 at 19:42

Dmitry S

4,990
2
24
32

Thanks @DanielVérité and Dmitry. – Peter Krauss Mar 04 '16 at 13:18
Do you know how to exclude stop words in a certain language? No stemming but language-based stop words – sparkle Oct 04 '19 at 12:29
If you want to avoid stopwords you could create a text search dictionary which simply used a blank file of stop words. For instance, see link for specifying a stopwords file in your dictionary. Then you'll need to create a text search configuration that uses that dictionary. https://www.postgresql.org/docs/current/textsearch-dictionaries.html#TEXTSEARCH-STOPWORDS – Peter Gerdes May 25 '22 at 02:50
Or better yet just directly skip them: https://stackoverflow.com/questions/1497895/can-i-configure-postgresql-programmatically-to-not-eliminate-stop-words-in-full – Peter Gerdes May 25 '22 at 03:06

score 0 · Answer 2 · answered May 25 '22 at 03:11

Just in case anyone else coming here is looking to really manually create a ts_vector you can skip using to_tsvector entierly (even with the simple dictionary it still sends it through the parser and drops spaces/punctuation e.g. ```to_tsvector('simple', '\Sigma^0_1') breaks that text up into three lexmes 0, 1 and sigma)

As described in this part of the manual and this answer. You simply do this

$$' space containing lexeme ' ' and one with position and weight'5c$$::tsvector;

How to use to_tsvector with no-dictionary (no stemming)?

2 Answers2