D-Net
/
dnet-hadoop
15
4
Fork 10
Code Issues 14 Pull Requests 8 Projects 2 Releases 1 Wiki Activity

[stats wf] final touches for the integration of PRs #166, #179 in the master branch

Browse Source
dump
Claudio Atzori 2 years ago
parent b053b0178e
commit 9acc32faa6
2 changed files with 29 additions and 224 deletions
Show Stats Download Patch File Download Diff File

236
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql
Unescape Escape View File

@ -42,181 +42,8 @@ join datasource on datasource.id = ri.hostedby
where datasource.id like '%doajarticles%') tmp
on p.id= tmp.id;
--create table indi_project_pubs_count stored as parquet as
--select pr.id id, count(p.id) total_pubs from project_results pr
--join publication p on p.id=pr.result
--group by pr.id;
--create table indi_project_datasets_count stored as parquet as
--select pr.id id, count(d.id) total_datasets from project_results pr
--join dataset d on d.id=pr.result
--group by pr.id;
--create table indi_project_software_count stored as parquet as
--select pr.id id, count(s.id) total_software from project_results pr
--join software s on s.id=pr.result
--group by pr.id;
--create table indi_project_otherresearch_count stored as parquet as
--select pr.id id, count(o.id) total_other from project_results pr
--join otherresearchproduct o on o.id=pr.result
--group by pr.id;
--create table indi_pub_avg_year_country_oa stored as parquet as
--select year, country, round(OpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageOA,
--round(NonOpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageNonOA
--from
--(SELECT year, country, SUM(CASE
--WHEN bestlicence='Open Access' THEN 1
--ELSE 0
--END) AS OpenAccess, SUM(CASE
--WHEN bestlicence<>'Open Access' THEN 1
--ELSE 0
--END) AS NonOpenAccess
--FROM publication p
--join result_organization ro on p.id=ro.id
--join organization o on o.id=ro.organization
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by year, country) tmp;
--create table indi_dataset_avg_year_country_oa stored as parquet as
--select year, country, round(OpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageOA,
--round(NonOpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageNonOA
--from
--(SELECT year, country, SUM(CASE
--WHEN bestlicence='Open Access' THEN 1
--ELSE 0
--END) AS OpenAccess, SUM(CASE
--WHEN bestlicence<>'Open Access' THEN 1
--ELSE 0
--END) AS NonOpenAccess
--FROM dataset d
--join result_organization ro on d.id=ro.id
--join organization o on o.id=ro.organization
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by year, country) tmp;
--create table indi_software_avg_year_country_oa stored as parquet as
--select year, country, round(OpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageOA,
--round(NonOpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageNonOA
--from
-- (SELECT year, country, SUM(CASE
--WHEN bestlicence='Open Access' THEN 1
-- ELSE 0
--END) AS OpenAccess, SUM(CASE
-- WHEN bestlicence<>'Open Access' THEN 1
-- ELSE 0
-- END) AS NonOpenAccess
-- FROM software s
-- join result_organization ro on s.id=ro.id
-- join organization o on o.id=ro.organization
-- where cast(year as int)>=2003 and cast(year as int)<=2021
-- group by year, country) tmp;
--create table indi_other_avg_year_country_oa stored as parquet as
--select year, country, round(OpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageOA,
--round(NonOpenAccess/(OpenAccess+NonOpenAccess)*100,3) as averageNonOA
-- from
-- (SELECT year, country, SUM(CASE
-- WHEN bestlicence='Open Access' THEN 1
-- ELSE 0
-- END) AS OpenAccess, SUM(CASE
-- WHEN bestlicence<>'Open Access' THEN 1
-- ELSE 0
-- END) AS NonOpenAccess
-- FROM otherresearchproduct orp
-- join result_organization ro on orp.id=ro.id
-- join organization o on o.id=ro.organization
-- where cast(year as int)>=2003 and cast(year as int)<=2021
-- group by year, country) tmp;
--create table indi_pub_avg_year_context_oa stored as parquet as
--with total as
--(select count(distinct pc.id) no_of_pubs, year, c.name name, sum(count(distinct pc.id)) over(PARTITION by year) as total from publication_concepts pc
--join context c on pc.concept like concat('%',c.id,'%')
--join publication p on p.id=pc.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by c.name, year )
--select year, name, round(no_of_pubs/total*100,3) averageofpubs
--from total;
--create table indi_dataset_avg_year_context_oa stored as parquet as
--with total as
--(select count(distinct pc.id) no_of_pubs, year, c.name name, sum(count(distinct pc.id)) over(PARTITION by year) as total from dataset_concepts pc
--join context c on pc.concept like concat('%',c.id,'%')
--join dataset p on p.id=pc.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by c.name, year )
--select year, name, round(no_of_pubs/total*100,3) averageofdataset
--from total;
--create table indi_software_avg_year_context_oa stored as parquet as
--with total as
--(select count(distinct pc.id) no_of_pubs, year, c.name name, sum(count(distinct pc.id)) over(PARTITION by year) as total from software_concepts pc
--join context c on pc.concept like concat('%',c.id,'%')
--join software p on p.id=pc.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by c.name, year )
--select year, name, round(no_of_pubs/total*100,3) averageofsoftware
--from total;
--create table indi_other_avg_year_context_oa stored as parquet as
--with total as
--(select count(distinct pc.id) no_of_pubs, year, c.name name, sum(count(distinct pc.id)) over(PARTITION by year) as total from otherresearchproduct_concepts pc
--join context c on pc.concept like concat('%',c.id,'%')
--join otherresearchproduct p on p.id=pc.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by c.name, year )
--select year, name, round(no_of_pubs/total*100,3) averageofother
--from total;
--create table indi_other_avg_year_content_oa stored as parquet as
--with total as
--(select count(distinct pd.id) no_of_pubs, year, d.type type, sum(count(distinct pd.id)) over(PARTITION by year) as total
--from otherresearchproduct_datasources pd
--join datasource d on datasource=d.id
--join otherresearchproduct p on p.id=pd.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by d.type, year)
--select year, type, round(no_of_pubs/total*100,3) averageOfOtherresearchproduct
--from total;
--create table indi_software_avg_year_content_oa stored as parquet as
--with total as
--(select count(distinct pd.id) no_of_pubs, year, d.type type, sum(count(distinct pd.id)) over(PARTITION by year) as total
--from software_datasources pd
--join datasource d on datasource=d.id
--join software p on p.id=pd.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by d.type, year)
--select year, type, round(no_of_pubs/total*100,3) averageOfSoftware
--from total;
--create table indi_dataset_avg_year_content_oa stored as parquet as
--with total as
--(select count(distinct pd.id) no_of_pubs, year, d.type type, sum(count(distinct pd.id)) over(PARTITION by year) as total
--from dataset_datasources pd
--join datasource d on datasource=d.id
--join dataset p on p.id=pd.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by d.type, year)
--select year, type, round(no_of_pubs/total*100,3) averageOfDatasets
--from total;
--create table indi_pub_avg_year_content_oa stored as parquet as
--with total as
--(select count(distinct pd.id) no_of_pubs, year, d.type type, sum(count(distinct pd.id)) over(PARTITION by year) as total
--from publication_datasources pd
--join datasource d on datasource=d.id
--join publication p on p.id=pd.id
--where cast(year as int)>=2003 and cast(year as int)<=2021
--group by d.type, year)
--select year, type, round(no_of_pubs/total*100,3) averageOfPubs
--from total;
---- Sprint 2 ----
create table indi_result_has_cc_licence_f stored as parquet as
create table indi_result_has_cc_licence stored as parquet as
select distinct r.id, (case when lic='' or lic is null then 0 else 1 end) as has_cc_license
from result r
left outer join (select r.id, license.type as lic from result r
@ -233,35 +60,6 @@ join result_licenses as license on license.id = r.id
WHERE lower(parse_url(license.type, "HOST")) = "creativecommons.org") tmp
on r.id= tmp.id;
-- EOSC-TR1.1-02M:
-- ## Indicator: has_cc_license. Creative Commons licensing has become a
-- de facto standard in scholarly communication and is promoted by many initiatives
-- like Plan S. This indicator might be only useful when applied
-- to openly available publications.
--create table indi_pub_has_cc_licence_tr stored as parquet as
--select distinct p.id, case when lic='' or lic is null then 0 else 1 end indi_result_org_collabas has_cc_license_tr
--from publication p
--left outer join (select p.id, license.type as lic from publication p
--join publication_licenses as license on license.id = p.id
--where lower(license.type) LIKE '%creativecommons.org%' OR lower(license.type) LIKE '%cc-%') tmp
--on p.id= tmp.id
-- #EOSC-F2-01M_cc Rich metadata for scholarly publications
-- ## Indicator: has_cc_license. Creative Commons licensing has become a
-- de facto standard in scholarly communication and is promoted by many initiatives
-- like Plan S. This indicator might be only useful when applied
-- to openly available publications.
-- Same indicator as EOSC-TR1.1-02M (Najko's instructions)
-- create table indi_pub_has_cc_licence_f stored as parquet as
-- select
-- distinct p.id, case when lic='' or lic is null then 0 else 1 end as has_cc_license_f
-- from publication p
-- left outer join (selectp.id,license.type as lic from publication p
-- join publication_licenses as license on license.id = p.id
-- where lower(license.type) LIKE '%creativecommons.org%' OR lower(license.type) LIKE '%cc-%') tmp
-- on p.id= tmp.id
create table indi_pub_has_abstract stored as parquet as
select distinct publication.id, coalesce(abstract, 1) has_abstract
from publication;
@ -270,16 +68,16 @@ create table indi_result_with_orcid stored as parquet as
select distinct r.id, coalesce(has_orcid, 0) as has_orcid
from result r
left outer join (select id, 1 as has_orcid from result_orcid) tmp
on r.id= tmp.id
on r.id= tmp.id;
---- Sprint 3 ----
---- Sprint 3 ----
create table indi_funded_result_with_fundref stored as parquet as
select distinct r.id, coalesce(fundref, 0) as fundref
from project_results r
left outer join (select distinct id, 1 as fundref from project_results
where provenance='Harvested') tmp
on r.id= tmp.id
on r.id= tmp.id;
create table indi_result_org_country_collab stored as parquet as
with tmp as
@ -290,7 +88,7 @@ select o1.id org1,o2.country country2, o1.type, count(distinct o1.result) as col
from tmp as o1
join tmp as o2 on o1.result=o2.result
where o1.id<>o2.id and o1.country<>o2.country
group by o1.id, o1.type,o2.country
group by o1.id, o1.type,o2.country;
create table indi_result_org_collab stored as parquet as
with tmp as
@ -301,18 +99,18 @@ select o1.id org1,o2.id org2, o1.type, count(distinct o1.result) as collaboratio
from tmp as o1
join tmp as o2 on o1.result=o2.result
where o1.id<>o2.id
group by o1.id, o2.id, o1.type
group by o1.id, o2.id, o1.type;
create table indi_funder_country_collab stored as parquet as
with tmp as (select funder, project, country from organization_projects op
join organization o on o.id=op.id
join project p on p.id=op.project
where country <> 'UNKNOWN')
select f1.funder, f1.country, f2.country, count(distinct f1.project) as collaborations
select f1.funder, f1.country as country1, f2.country as country2, count(distinct f1.project) as collaborations
from tmp as f1
join tmp as f2 on f1.project=f2.project
where f1.country<>f2.country
group by f1.funder, f2.country, f1.country
group by f1.funder, f2.country, f1.country;
create table indi_result_country_collab stored as parquet as
with tmp as
@ -323,10 +121,9 @@ select o1.country country1, o2.country country2, o1.type, count(distinct o1.resu
from tmp as o1
join tmp as o2 on o1.result=o2.result
where o1.country<>o2.country
group by o1.country, o2.country, o1.type
group by o1.country, o2.country, o1.type;
---- Sprint 4 ----
create table indi_pub_diamond stored as parquet as
select distinct pd.id, coalesce(in_diamond_journal, 0) as in_diamond_journal
from publication_datasources pd
@ -335,7 +132,7 @@ select pd.id, 1 as in_diamond_journal from publication_datasources pd
join datasource d on d.id=pd.datasource
join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and (ps.journal_is_in_doaj=true or ps.journal_is_oa=true) and ps.has_apc=false) tmp
on pd.id=tmp.id
on pd.id=tmp.id;
create table indi_pub_hybrid stored as parquet as
select distinct pd.id, coalesce(is_hybrid, 0) as is_hybrid
@ -345,7 +142,7 @@ select pd.id, 1 as is_hybrid from publication_datasources pd
join datasource d on d.id=pd.datasource
join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and (ps.journal_is_in_doaj=false and ps.journal_is_oa=false)) tmp
on pd.id=tmp.id
on pd.id=tmp.id;
create table indi_is_gold_oa stored as parquet as
(select distinct pd.id, coalesce(gold_oa, 0) as gold_oa
@ -355,8 +152,7 @@ select pd.id, 1 as gold_oa from publication_datasources pd
join datasource d on d.id=pd.datasource
join stats_ext.plan_s_jn ps on (ps.issn_print=d.issn_printed or ps.issn_online=d.issn_online)
where ps.journal_is_in_doaj is true or ps.journal_is_oa is true) tmp
on pd.id=tmp.id)
on pd.id=tmp.id);
create table indi_pub_in_transformative stored as parquet as
select distinct pd.id, coalesce(is_transformative, 0) as is_transformative
@ -366,7 +162,7 @@ select pd.id, 1 as is_transformative from publication_datasources pd
join datasource d on d.id=pd.datasource
join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and ps.is_transformative_journal=true) tmp
on pd.id=tmp.id
on pd.id=tmp.id;
create table indi_pub_closed_other_open stored as parquet as
select distinct ri.id, coalesce(pub_closed_other_open, 0) as pub_closed_other_open from result_instance ri
@ -376,9 +172,9 @@ join publication p on p.id=ri.id
join datasource d on ri.hostedby=d.id
where d.type like '%Journal%' and ri.accessright='Closed Access' and
(p.bestlicence='Open Access' or p.bestlicence='Open Source')) tmp
on tmp.id=ri.id
on tmp.id=ri.id;
---- Sprint 5 ----
---- Sprint 5 ----
create table indi_result_no_of_copies stored as parquet as
select id, count(id) as number_of_copies from result_instance group by id
select id, count(id) as number_of_copies from result_instance group by id;

17
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql
Unescape Escape View File

@ -49,8 +49,10 @@ compute stats TARGET.result_greenoa;
create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_languages;
create table TARGET.result_licences stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_licences;
create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_licenses;
create table TARGET.licenses_normalized as select * from SOURCE.licenses_normalized;
create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_oids;
@ -79,6 +81,9 @@ compute stats TARGET.result_sources;
create table TARGET.result_topics stored as parquet as select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_topics;
create table TARGET.result_result stored as parquet as select * from SOURCE.result_result orig where exists (select 1 from TARGET.result r where r.id=orig.source or r.id=orig.target);
compute stats TARGET.result_result;
-- datasources
create view if not exists TARGET.datasource as select * from SOURCE.datasource;
create view if not exists TARGET.datasource_oids as select * from SOURCE.datasource_oids;
@ -100,6 +105,7 @@ create view if not exists TARGET.project as select * from SOURCE.project;
create view if not exists TARGET.project_oids as select * from SOURCE.project_oids;
create view if not exists TARGET.project_organizations as select * from SOURCE.project_organizations;
create view if not exists TARGET.project_resultcount as select * from SOURCE.project_resultcount;
create view if not exists TARGET.project_classification as select * from SOURCE.project_classification;
create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects;
compute stats TARGET.project_results;
@ -115,8 +121,8 @@ create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.i
compute stats TARGET.indi_pub_gold_oa;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_pub_has_abstract;
create table TARGET.indi_result_has_cc_licence_f stored as parquet as select * from SOURCE.indi_result_has_cc_licence_f orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_result_has_cc_licence_f;
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_result_has_cc_licence;
create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_result_has_cc_licence_url;
@ -138,6 +144,9 @@ compute stats TARGET.indi_pub_closed_other_open;
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_result_no_of_copies;
--- Usage statistics
create table TARGET.usage_stats stored as parquet as select * from SOURCE.usage_stats orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--denorm
alter table TARGET.result rename to TARGET.res_tmp;

Write Preview
Loading…
Cancel
Save