Add new version of the dataset

pln-fing-udelar · May 27, 2018 · 3146889 · 3146889
1 parent bd545e6
commit 3146889
Show file tree

Hide file tree

Showing 14 changed files with 324,754 additions and 99,891 deletions.
diff --git a/annotations.csv b/annotations.csv
diff --git a/annotations_by_tweet.csv b/annotations_by_tweet.csv
diff --git a/annotations_by_tweet_all.csv b/annotations_by_tweet_all.csv
diff --git a/annotations_by_tweet_top.csv b/annotations_by_tweet_top.csv
diff --git a/extraction/analyses.sql b/extraction/analyses.sql
@@ -139,7 +139,7 @@ FROM tweets t
                 GROUP BY session_id
                 HAVING COUNT(*) > 1000) top_sessions
 GROUP BY t.tweet_id
-INTO OUTFILE '/var/lib/mysql-files/annotations_by_tweet_top2.csv' FIELDS TERMINATED BY ','
+INTO OUTFILE '/var/lib/mysql-files/annotations_by_tweet_top.csv' FIELDS TERMINATED BY ','
   OPTIONALLY ENCLOSED BY '"'
   ESCAPED BY '' LINES TERMINATED BY '\n';
 
@@ -206,3 +206,15 @@ FROM tweets
 INTO OUTFILE '/var/lib/mysql-files/tweets.csv' FIELDS TERMINATED BY ','
   OPTIONALLY ENCLOSED BY '"'
   ESCAPED BY '' LINES TERMINATED BY '\n';
+-- Then add the CSV headers manually.
+
+SELECT
+  tweet_id,
+  session_id,
+  date,
+  vote
+FROM votes
+INTO OUTFILE '/var/lib/mysql-files/annotations.csv' FIELDS TERMINATED BY ','
+  OPTIONALLY ENCLOSED BY'"'
+  ESCAPED BY '' LINES TERMINATED BY '\n';
+-- Then add the CSV headers manually.
diff --git a/index.html b/index.html
@@ -109,7 +109,7 @@ <h2>Download</h2>
 <h2>Roadmap</h2>
 
 <ol>
-  <li>Add more information to this website - ETA: Dec 2017</li>
+  <li>Add and update information in this website (there's old stuff in here!) - no ETA yet</li>
   <li>HUMOR dataset bundled in NLTK - no ETA yet</li>
 </ol>
 

diff --git a/previous/annotations.csv b/previous/annotations.csv
diff --git a/previous/annotations_by_tweet.csv b/previous/annotations_by_tweet.csv
diff --git a/annotations_by_tweet_to_csv.sql → previous/annotations_by_tweet_to_csv.sql b/annotations_by_tweet_to_csv.sql → previous/annotations_by_tweet_to_csv.sql
diff --git a/annotations_to_csv.sql → previous/annotations_to_csv.sql b/annotations_to_csv.sql → previous/annotations_to_csv.sql
diff --git a/corpus-v1.sql → previous/corpus-v1.sql b/corpus-v1.sql → previous/corpus-v1.sql