Merge pull request #398 from georgetown-cset/348-normalize-yearly-cou…

…nt-column-names

Standardize yearly count column names

company_linkage/sql/all_paper_counts.sql

@@ -1,7 +1,7 @@
 WITH year_counts AS (
   SELECT
     CSET_id,
-    COUNT(DISTINCT(merged_id)) as year_count,
+    COUNT(DISTINCT(merged_id)) as num_papers,
     year
   FROM
     staging_ai_companies_visualization.all_company_papers
@@ -10,8 +10,8 @@ WITH year_counts AS (
 
 SELECT
   CSET_id,
-  SUM(year_count) as all_pubs,
-  ARRAY_AGG(STRUCT(year, year_count)) AS all_pubs_by_year
+  SUM(num_papers) as all_pubs,
+  ARRAY_AGG(STRUCT(year, num_papers)) AS all_pubs_by_year
 FROM
   year_counts
 GROUP BY

company_linkage/sql/highly_cited_paper_counts.sql

@@ -1,7 +1,7 @@
 WITH year_counts AS (
   SELECT
     CSET_id,
-    COUNT(DISTINCT(merged_id)) as year_count,
+    COUNT(DISTINCT(merged_id)) as num_papers,
     year
   FROM
     staging_ai_companies_visualization.highly_cited_ai_publications
@@ -10,8 +10,8 @@ WITH year_counts AS (
 
 SELECT
   CSET_id,
-  SUM(year_count) as highly_cited_ai_pubs,
-  ARRAY_AGG(STRUCT(year, year_count)) AS highly_cited_ai_pubs_by_year
+  SUM(num_papers) as highly_cited_ai_pubs,
+  ARRAY_AGG(STRUCT(year, num_papers)) AS highly_cited_ai_pubs_by_year
 FROM
   year_counts
 GROUP BY

company_linkage/sql/initial_paper_visualization_data.sql

@@ -47,19 +47,19 @@ all_cited as
 (SELECT
   CSET_id,
   ARRAY_AGG(STRUCT(year,
-      ai_citation_count)
+      ai_citation_count as num_papers)
   ORDER BY
     year) AS ai_citation_count_by_year,
   ARRAY_AGG(STRUCT(year,
-      cv_citation_count)
+      cv_citation_count as num_papers)
   ORDER BY
     year) AS cv_citation_count_by_year,
   ARRAY_AGG(STRUCT(year,
-      nlp_citation_count)
+      nlp_citation_count as num_papers)
   ORDER BY
     year) AS nlp_citation_count_by_year,
   ARRAY_AGG(STRUCT(year,
-      robotics_citation_count)
+      robotics_citation_count as num_papers)
   ORDER BY
     year) AS robotics_citation_count_by_year
 FROM

company_linkage/sql/patent_visualization_data_with_all_patents.sql

@@ -28,7 +28,7 @@ WITH
   SELECT
     CSET_id,
     ARRAY_AGG(STRUCT(year,
-        all_patents)
+        all_patents as num_patents)
     ORDER BY
       year) AS all_patents_by_year,
   FROM

company_linkage/sql/patent_visualization_data_with_by_year.sql

@@ -55,150 +55,149 @@ WITH
   SELECT
     CSET_id,
     ARRAY_AGG(STRUCT(priority_year,
-        ai_patents)
+        ai_patents as num_patents)
     ORDER BY
       priority_year) AS ai_patents_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Physical_Sciences_and_Engineering_pats)
+        Physical_Sciences_and_Engineering_pats as num_patents)
     ORDER BY
       priority_year) AS Physical_Sciences_and_Engineering_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Life_Sciences_pats)
+        Life_Sciences_pats as num_patents)
     ORDER BY
       priority_year) AS Life_Sciences_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Security__eg_cybersecurity_pats)
+        Security__eg_cybersecurity_pats as num_patents)
     ORDER BY
       priority_year) AS Security__eg_cybersecurity_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Transportation_pats)
+        Transportation_pats as num_patents)
     ORDER BY
       priority_year) AS Transportation_pats_by_year,
       ARRAY_AGG(STRUCT(priority_year,
-        Industrial_and_Manufacturing_pats)
+        Industrial_and_Manufacturing_pats as num_patents)
     ORDER BY
       priority_year) AS Industrial_and_Manufacturing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Education_pats)
+        Education_pats as num_patents)
     ORDER BY
       priority_year) AS Education_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Document_Mgt_and_Publishing_pats)
+        Document_Mgt_and_Publishing_pats as num_patents)
     ORDER BY
       priority_year) AS Document_Mgt_and_Publishing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Military_pats)
+        Military_pats as num_patents)
     ORDER BY
       priority_year) AS Military_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Agricultural_pats)
+        Agricultural_pats as num_patents)
     ORDER BY
       priority_year) AS Agricultural_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Computing_in_Government_pats)
+        Computing_in_Government_pats as num_patents)
     ORDER BY
       priority_year) AS Computing_in_Government_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Personal_Devices_and_Computing_pats)
+        Personal_Devices_and_Computing_pats as num_patents)
     ORDER BY
       priority_year) AS Personal_Devices_and_Computing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Banking_and_Finance_pats)
+        Banking_and_Finance_pats as num_patents)
     ORDER BY
       priority_year) AS Banking_and_Finance_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Telecommunications_pats)
+        Telecommunications_pats as num_patents)
     ORDER BY
       priority_year) AS Telecommunications_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Networks__eg_social_IOT_etc_pats)
+        Networks__eg_social_IOT_etc_pats as num_patents)
     ORDER BY
       priority_year) AS Networks__eg_social_IOT_etc_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Business_pats)
+        Business_pats as num_patents)
     ORDER BY
       priority_year) AS Business_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Energy_Management_pats)
+        Energy_Management_pats as num_patents)
     ORDER BY
       priority_year) AS Energy_Management_pats_by_year,
       ARRAY_AGG(STRUCT(priority_year,
-        Entertainment_pats)
+        Entertainment_pats as num_patents)
     ORDER BY
       priority_year) AS Entertainment_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Nanotechnology_pats)
+        Nanotechnology_pats as num_patents)
     ORDER BY
       priority_year) AS Nanotechnology_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Semiconductors_pats)
+        Semiconductors_pats as num_patents)
     ORDER BY
       priority_year) AS Semiconductors_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Language_Processing_pats)
+        Language_Processing_pats as num_patents)
     ORDER BY
       priority_year) AS Language_Processing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Speech_Processing_pats)
+        Speech_Processing_pats as num_patents)
     ORDER BY
       priority_year) AS Speech_Processing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Knowledge_Representation_pats)
+        Knowledge_Representation_pats as num_patents)
     ORDER BY
       priority_year) AS Knowledge_Representation_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Planning_and_Scheduling_pats)
+        Planning_and_Scheduling_pats as num_patents)
     ORDER BY
       priority_year) AS Planning_and_Scheduling_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Control_pats)
+        Control_pats as num_patents)
     ORDER BY
       priority_year) AS Control_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Distributed_AI_pats)
+        Distributed_AI_pats as num_patents)
     ORDER BY
       priority_year) AS Distributed_AI_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Robotics_pats)
+        Robotics_pats as num_patents)
     ORDER BY
       priority_year) AS Robotics_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Computer_Vision_pats)
+        Computer_Vision_pats as num_patents)
     ORDER BY
       priority_year) AS Computer_Vision_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Analytics_and_Algorithms_pats)
+        Analytics_and_Algorithms_pats as num_patents)
     ORDER BY
       priority_year) AS Analytics_and_Algorithms_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Measuring_and_Testing_pats)
+        Measuring_and_Testing_pats as num_patents)
     ORDER BY
       priority_year) AS Measuring_and_Testing_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Logic_Programming_pats)
+        Logic_Programming_pats as num_patents)
     ORDER BY
       priority_year) AS Logic_Programming_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Fuzzy_Logic_pats)
+        Fuzzy_Logic_pats as num_patents)
     ORDER BY
       priority_year) AS Fuzzy_Logic_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Probabilistic_Reasoning_pats)
+        Probabilistic_Reasoning_pats as num_patents)
     ORDER BY
       priority_year) AS Probabilistic_Reasoning_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Ontology_Engineering_pats)
+        Ontology_Engineering_pats as num_patents)
     ORDER BY
       priority_year) AS Ontology_Engineering_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Machine_Learning_pats)
+        Machine_Learning_pats as num_patents)
     ORDER BY
       priority_year) AS Machine_Learning_pats_by_year,
     ARRAY_AGG(STRUCT(priority_year,
-        Search_Methods_pats)
+        Search_Methods_pats as num_patents)
     ORDER BY
       priority_year) AS Search_Methods_pats_by_year,
-
   FROM
     high_resolution_entities.aggregated_organizations
   LEFT JOIN