change citation formatting

pages/2_Dataset_Statistics.py

@@ -7,9 +7,6 @@ from filter_dataframe import filter_dataframe
 def get_language_stats_df():
     return pd.read_parquet("data/datasets_stats.parquet")
 
-def export_citations(df: pd.DataFrame) -> str:
-    return "\n\n".join(df["citation"].tolist())
-
 _MMS_CITATION = """\
 @misc{augustyniak2023massively,
       title={Massively Multilingual Corpus of Sentiment Datasets and Multi-faceted Sentiment Classification Benchmark}, 
@@ -20,7 +17,7 @@ _MMS_CITATION = """\
       primaryClass={cs.CL}
 }"""
 
-CITATION_SEPARATOR = "\n% " + ("-" * 90) + "\n"
+CITATION_SEPARATOR = "% " + ("-" * 90) + "\n\n"
 
 def export_citations(df: pd.DataFrame):
     dataset_names = df.original_dataset.tolist()
@@ -30,7 +27,7 @@ def export_citations(df: pd.DataFrame):
     citations_grouped_df = df.groupby("citation").agg({"dataset": lambda x: ", ".join(x)}).reset_index().sort_values(by="dataset")
     dataset_citations = ("% Datasets: " + citations_grouped_df["dataset"] + "\n" + citations_grouped_df["citation"]).to_list()
     dataset_citations_joined = CITATION_SEPARATOR.join(dataset_citations)
-    return f"{_MMS_CITATION}\n\n{dataset_citations_joined}"
+    return f"% MMS corpus citation\n{_MMS_CITATION}\n{CITATION_SEPARATOR}{dataset_citations_joined}"
 
 
 st.set_page_config(page_title="Dataset statistics", page_icon="📈")