Upload dashboard.py

dashboard.py

@@ -1,4 +1,7 @@
 import warnings
+
+from bokeh.models import DatetimeTicker
+
 warnings.filterwarnings("ignore")
 import io
 import os
@@ -154,11 +157,12 @@ grouping_filtered = pd.read_csv(os.path.join(data_folder, 'dna_relations.tsv'),
 def customize_x_axis_bokeh(plot, element):
     bokeh_plot = plot.state
     bokeh_plot.xaxis.major_label_orientation = 45  # Rotate x-axis labels
+    bokeh_plot.xaxis.ticker = DatetimeTicker(desired_num_ticks=6)
 
 def create_publication_curve_chart():
     country_name_df = pd.read_csv(os.path.join(data_folder, 'country_name_map.tsv'),  header=0, sep='\t', lineterminator='\n', low_memory=False)
     country_name_map = dict(zip(country_name_df.Country_Code, country_name_df.Country_Name))
-    country_name_map
+    #country_name_map
     total_publications_time_indexed = pd.read_csv(os.path.join(data_folder, 'total_publications_time_indexed.tsv'),  header=0, sep='\t', lineterminator='\n', low_memory=False)
     country_publications_time_indexed = pd.read_csv(os.path.join(data_folder, 'country_publications_time_indexed.tsv'),  header=0, sep='\t', lineterminator='\n', low_memory=False)
     total_publications_time_indexed.id = np.log1p(total_publications_time_indexed.id)
@@ -169,7 +173,7 @@ def create_publication_curve_chart():
     curve_countries = []
     for country in country_name_map.keys():
        overlay = overlay * hv.Curve((total_publications_time_indexed.month_bin, country_publications_time_indexed[country]), label=country_name_map[country])
-    overlay.opts(show_legend=True,legend_position='right', width=1200, height=500, hooks=[customize_x_axis_bokeh])
+    overlay.opts(show_legend=True,legend_position='right', width=1400, height=600, hooks=[customize_x_axis_bokeh])
     return overlay
 
 
@@ -374,7 +378,7 @@ def generate_radio_buttons(value):
 button0 = pn.widgets.Button(name="Introduction", button_type="warning", icon="file-info", styles={"width": "100%"})
 button1 = pn.widgets.Button(name="Publication Trends", button_type="warning", icon="chart-histogram", styles={"width": "100%"})
 button2 = pn.widgets.Button(name="Topic Map", button_type="warning", icon="chart-dots-3", styles={"width": "100%"})
-button3 = pn.widgets.Button(name="AECO Macro Topics Hierarchy", button_type="warning",  icon="chart-dots-3", styles={"width": "100%"})
+button3 = pn.widgets.Button(name="AECO Macro Topic Hierarchy", button_type="warning",  icon="chart-dots-3", styles={"width": "100%"})
 button4 = pn.widgets.Button(name="AECO Macro Topics Trends", button_type="warning", icon="chart-histogram", styles={"width": "100%"})
 button5 = pn.widgets.Button(name="Research Collaboration Networks: Institutes", button_type="warning", icon="chart-dots-3", styles={"width": "100%"})
 button6 = pn.widgets.Button(name="Research Collaboration Networks: Countries", button_type="warning", icon="chart-dots-3", styles={"width": "100%"})
@@ -435,23 +439,49 @@ button6.on_click(lambda event: show_page("Page6"))
 def CreatePage0():
     return pn.Column(pn.pane.Markdown("""
 
-This is a dashboard for a Research Analysis project regarding research and technology in the AECO domain. The source data consists of around 
-276k English-language research papers gathered from the openalex.org graph database, covering a timeframe from 2011 through 2024.
+
+## Introduction
+
+This is a dashboard for a Data Analytics project regarding research publications in the AECO domain. The source data consists of around 
+267k English-language research papers gathered from the openalex.org graph database, covering a timeframe from 2011 through early 2024.
 
 ---------------------------
 
-## AECO Topic Map
-In the AECO Topic Map panel we show  the 6-month-sampled time series depicting the number of published research papers 
-for the 16 macro-topics automatically detected by an optimized BerTopic model and ppst-processed for manual topic merging.
+## Publication Trends
+
+In the "Publication Trends" panel we show monthly time series of the total number of publications and the number of publications per country (both in log scale), for the top 20 
+countries by number of publications in the dataset.
+
+
+## Topic Map
+In the "Topic Map" panel we show a UMAP reduced 2-dimensional visualization of the optimized 52 topic clusters of AECO research papers, embedded using Sentence Transformer model, 
+with the descriptive labels overlayed on the clusters being generated by LLama 2 Large Language Model. Each point in the space represent a paper from the dataset. 
+Hovering over it has the paper title popping up, while clicking on it redirects to the corresponding OpenAlex paper entry page.
+ 
+ 
+## AECO Macro Topic Hierarchy
+The "AECO Macro Topic Hierarchy" panel allows to explore the dendrogram representation of the optimized clustering, with the leaves of the tree
+representing the 51 clusters, the intermediate nodes representing merged clusters and the height of the merging (distance from the leaves) indicating topic
+similarity as based on the cosine distance matrix between topic embeddings.
+ 
+ 
+## AECO Macro Topic Trends
+The "AECO Macro Topic Trends" panel shows the evolution over time of the 16 AECO macro topics by plotting the semi-annual time series of the absolute numbers of publications per topic.
+
+## Research Collaboration Networks: Institutes
+The "Research Collaboration Networks: Institutes" panel contains VOSViewer-generated network representations of the research institutions co-authorship connections, for each of the 16 macro
+clusters. By selecting a macro-cluster from by the drop-down menu, a graph is loaded whose nodes represent research institutions, edges represent co-authorships relations (with edge 
+thickness being proportional to the frequency of the relations) and the color code clustering highlights the partition of the graph in highly interconnected node groups.
+Open the VOSViewer left panel to customize the visualization and/or search for a target institution in the graph.
 
-## AECO Macro Topics
-In the AECO Macro Topics panel we present the 6-month-sampled time series depicting the number of published research papers 
-for the 16 macro-topics automatically detected by an optimized BerTopic model and ppst-processed for manual topic merging.
 
+## Research Collaboration Networks: Countries
+The "Research Collaboration Networks: Countries" panel contains VOSViewer-generated network representations of the authors/institutions' country co-authorship connections, for each of 
+the 16 macro clusters. By selecting a macro-cluster from by the drop-down menu, a graph is loaded whose nodes represent authors/institutions' countries, edges represent co-authorships 
+relations (with edge thickness being proportional to the frequency of the relations) and the color code clustering highlights the partition of the graph in highly interconnected node
+ groups. Open the VOSViewer left panel to customize the visualization and/or search for a target country in the graph.
 
-### Research Collaboration Networks: Institutes
 
-### Research Collaboration Networks: Authors
 """, width=800), align="center")
 
 def CreatePage1():
@@ -522,13 +552,13 @@ mapping = {
 }
 
 #################### SIDEBAR LAYOUT ##########################
-sidebar = pn.Column(pn.pane.Markdown("## Pages"),button0,button1,button2,button3,button4,button5,button6,
+sidebar = pn.Column(pn.pane.Markdown("## Panels"),button0,button1,button2,button3,button4,button5,button6,
                     #button5,
                     #button6,
                     styles={"width": "100%", "padding": "15px"})
 
 #################### MAIN AREA LAYOUT ##########################
-main_area = pn.Column(mapping["Page1"], styles={"width":"100%"})
+main_area = pn.Column(mapping["Page0"], styles={"width":"100%"})
 
 ###################### APP LAYOUT ##############################
 template = pn.template.BootstrapTemplate(