Add v7 datasets in datasets.py (#15)

asreview · Apr 14, 2020 · 35b0f83 · 35b0f83
1 parent ab11787
commit 35b0f83
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 20 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,3 +1,9 @@
+# R 
+
+.Rhistory
+
+# Python
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]

diff --git a/asreviewcontrib/covid19/datasets.py b/asreviewcontrib/covid19/datasets.py
@@ -9,39 +9,37 @@ class Cord19Dataset(BaseDataSet):
     topic = "Covid-19"
     license = "Covid dataset license"
     link = "https://pages.semanticscholar.org/coronavirus-research"
-    last_update = "2020-04-03"
+    last_update = "2020-04-10"
     description = "A free dataset on publications on the corona virus."
     img_url = ("https://pages.semanticscholar.org/hs-fs/hubfs/"
                "covid-image.png?width=300&name=covid-image.png")
     link = "https://pages.semanticscholar.org/coronavirus-research"
     year = 2020
 
 
-class Cord19DatasetV6(Cord19Dataset):
-    dataset_id = "cord19-v6"
-    title = "CORD-19 v6"
-    date = "2020-04-03"
+class Cord19DatasetV7(Cord19Dataset):
+    dataset_id = "cord19-v7"
+    title = "CORD-19 v7"
+    date = "2020-04-10"
     statistics = {
-        "n_papers": 47298,
+        "n_papers": 51078,
         "n_missing_titles": 158,
-        "n_missing_abstracts": 8250,
+        "n_missing_abstracts": 8726,
     }
-    url = ("https://ai2-semanticscholar-cord-19.s3-us-west-2.amazonaws.com/2020-04-03/metadata.csv")  # noqa
-    sha512 = ("5ba3738e603e2b23c403a46fb2620360415ba3419b09b071f5a5ca16a96422aa78a5456cba7abb18279b6510174273694961b72c28006620c7f28571125cfae2")  # noqa
+    url = ("https://ai2-semanticscholar-cord-19.s3-us-west-2.amazonaws.com/2020-04-10/metadata.csv")  # noqa
 
 
-class Cord19DatasetV6_Dec2019(Cord19Dataset):
-    dataset_id = "cord19-v6-2020"
-    title = "CORD-19 v5 since Dec. 2019"
-    last_update = "2020-04-03"
+class Cord19DatasetV7_Dec2019(Cord19Dataset):
+    dataset_id = "cord19-v7-2020"
+    title = "CORD-19 v7 since Dec. 2019"
+    last_update = "2020-04-10"
     statistics = {
-        "n_papers": 4774,
+        "n_papers": 5753,
         "n_missing_titles": 2,
-        "n_missing_abstracts": 1103,
+        "n_missing_abstracts": 1422,
     }
-    date = "2020-04-05"
-    url = ("https://raw.githubusercontent.com/asreview/asreview-covid19/master/datasets/cord19_v6_20191201.csv")  # noqa
-    sha512 = ("390c04b690abff2f824ed837367048308c6573032b2e45301056c3cfa7b04e6ea5b00ebd691e1af2066a820335bea1bdad240df736bd69be0b3c436fca629759")  # noqa
+    date = "2020-04-14"
+    url = ("https://raw.githubusercontent.com/asreview/asreview-covid19/master/datasets/cord19_v7_20191201.csv")  # noqa
 
 
 class Covid19DataGroup(BaseDataGroup):
@@ -50,6 +48,6 @@ class Covid19DataGroup(BaseDataGroup):
 
     def __init__(self):
         super(Covid19DataGroup, self).__init__(
-            Cord19DatasetV6(),
-            Cord19DatasetV6_Dec2019(),
+            Cord19DatasetV7(),
+            Cord19DatasetV7_Dec2019(),
         )