From b6222ec4af97f8b04d50431b0a5f713227e86dbc Mon Sep 17 00:00:00 2001
From: David Pomerenke <46022183+davidpomerenke@users.noreply.github.com>
Date: Sun, 21 Jul 2024 15:04:29 +0200
Subject: [PATCH] refactor(keyword_trend.py): refine keyword topic definitions

---
 .../media_impact_monitor/issue_keywords.yaml  | 12 ++++++----
 .../trends/keyword_trend.py                   | 22 +++++++++----------
 2 files changed, 19 insertions(+), 15 deletions(-)

diff --git a/backend-python/media_impact_monitor/issue_keywords.yaml b/backend-python/media_impact_monitor/issue_keywords.yaml
index 05d88d57..3a384218 100644
--- a/backend-python/media_impact_monitor/issue_keywords.yaml
+++ b/backend-python/media_impact_monitor/issue_keywords.yaml
@@ -15,15 +15,18 @@ activism:
   - mahnwache
   - hungerstreik
   - ziviler ungehorsam
-climate_science:
+climate_general:
   - klimawandel
   - klimaerwärmung
   - erderwärmung
   - klimaschutz
-  - klimagerechtigkeit
-  - klimapolitik
-  - klimaneutral*
+climate_science:
+  - klimaforsch*
+  - klimawissenschaft*
+  - erderwärmung
+  - ipcc
 climate_policy:
+  - klimapoliti*
   - klimaneutral*
   - klimaziel*
   - klimaschutzpaket
@@ -41,6 +44,7 @@ climate_policy:
   - neun-euro-ticket
   - vergesellschaftung
   - schuldenschnitt
+  - klimagerechtigkeit
 climate_urgency:
   - klimakrise
   - klimakatastrophe
diff --git a/backend-python/media_impact_monitor/trends/keyword_trend.py b/backend-python/media_impact_monitor/trends/keyword_trend.py
index ac414023..e7e424b5 100644
--- a/backend-python/media_impact_monitor/trends/keyword_trend.py
+++ b/backend-python/media_impact_monitor/trends/keyword_trend.py
@@ -56,19 +56,19 @@ def load_keywords():
 def topic_queries(media_source: str) -> dict[str, str]:
     keywords = load_keywords()
     keyword_queries = {
-        "science": xs(keywords["climate_science"], media_source),
-        "policy": xs(keywords["climate_policy"], media_source),
-        "urgency": xs(keywords["climate_urgency"], media_source),
-        "all_excl_activism": xs_without_ys(
-            keywords["climate_science"]
-            + keywords["climate_policy"]
-            + keywords["climate_urgency"],
-            keywords["activism"],
-            media_source,
-        ),
+        "climate policy": xs(keywords["climate_policy"], media_source),
+        "climate science": xs(keywords["climate_science"], media_source),
+        "climate crisis framing": xs(keywords["climate_urgency"], media_source),
+        # "all_excl_activism": xs_without_ys(
+        #     keywords["climate_science"]
+        #     + keywords["climate_policy"]
+        #     + keywords["climate_urgency"],
+        #     keywords["activism"],
+        #     media_source,
+        # ),
     }
     if media_source != "web_google":
-        keyword_queries["activism"] = xs_with_ys(
+        keyword_queries["climate activism"] = xs_with_ys(
             keywords["climate_science"]
             + keywords["climate_policy"]
             + keywords["climate_urgency"],