"SDG" phrase - Githubissues

SDGforskning / sdg-strings

Search strings for finding SDG-related works per goal and target (python script, WoS search strategy)

Creative Commons Attribution 4.0 International

3 stars 4 forks source link

"SDG" phrase #3

Closed Caroline-A closed 2 years ago

Caroline-A commented 3 years ago

I'm unsure what to do with the string that finds results actually mentioning "SDG 3". Variations give very different results. Is it enough to just mention the SDGs?

Maybe we can discuss this later, when we have all worked on all the strings - this issue is just here to remind us to go back and look at it before the final version.

EliSeland commented 2 years ago

SDG is short for "SET DOMAIN GROUP" - should perhaps be excluded with a NOT

A thought: perhaps the general SDG strings belong in the topic searches, but not the action searches?

EliSeland commented 2 years ago

We decided in a project group meeting on 27 April 2022 to remove the general SDG strings from the Action approach, and to use them in the Topic approach.

Caroline-A commented 2 years ago

[x] SDG 1 removed
[x] SDG 2 removed
[x] SDG 3 removed
[x] SDG 4 removed
[x] SDG 7 removed
[x] SDG 11 removed
[x] SDG 13 removed
[x] SDG 14 removed

Caroline-A commented 2 years ago

Reopened now that we are working on the topic approach

Caroline-A commented 2 years ago

They still some work, as a few problems have been identified and need further looking into:

SDG can be short/acronym for other things/terms. Some terms that cause noise are: Set Domain Group (identified in SDG 1,2), Spleen-Derived Growth factor (SDG 3), Steel Design Guide series (SDG 2), Secoisolariciresinol diglucoside (SDG 7). These should be considered excluded with a NOT, and we may need to try and identify more.
Numbers sometimes behave in particular ways. The search for SDG 4 for instance returns hits with SDG 3.4, SDG 7 for SDG 4.7 etc. - I managed to recreate this issue. I think searching for "SDG 4" OR "SDGs 4" is ok, but searching for "SDG$ 4" can create this issue, e.g. https://www.webofscience.com/wos/woscc/summary/0bc7400a-752c-47d0-b942-69b2b2f8c665-4be3d473/relevance/1
The key words included in the general search strings may need some adjustment. For instance "cities" causes noise in SDG 11, "health" may be too wide in SDG 3 etc.
Quite a few works list SDGs, e.g. "this work looks at SDGs to do with people (SDGs 1, 3, 4)". How to catch these?
"SDG 4" search seems to ignore punctuation, i.e. finds (SDG) #4 (SDG) 4 SDG#4 SDG-4 SDG 4.1

Caroline-A commented 2 years ago

The basic string we originally used was this (with the above modification/removal of SDG$). This seems to work relatively well for SDG 1, but will likely need modification for some others.

TS=
(
  "SDG 1" OR "SDGs 1" OR "SDG1" OR "sustainable development goal$ 1"
    OR
    (
      ("sustainable development goal$" OR "SDG$" OR "goal 1")
      NEAR/15 "poverty"
    )
)

Caroline-A commented 2 years ago

I've now added a suggestion to all the topic-level files