updates

requirements.txt

@@ -10,4 +10,5 @@ torch==1.12.1
 yellowbrick==1.5
 transformers==4.22.1
 textdistance==4.5.0
-datasets==2.5.2
+datasets==2.5.2
+bokeh==2.4.1

widgets/__init__.py

@@ -1,3 +1,3 @@
 from .body import render_body
 from .sidebar import render_sidebar
-from .utils import readfile, generate_html_pyecharts
+from .utils import readfile, generate_html_pyecharts

widgets/body.py

@@ -5,6 +5,7 @@ from pyecharts.charts import Bar
 from pyecharts import options as opts
 import streamlit.components.v1 as st_render
 from .utils import generate_html_pyecharts
+from .charts import build_bar_charts
 
 def __preview__(platforms, num_papers, num_papers_preview, query_input,start_year,end_year):
     with st.spinner('Searching...'):
@@ -36,7 +37,7 @@ We have found following papers for you! (displaying 5 papers for each literature
                 title = str(arxiv[i]['title']).replace('\n', ' ')
                 publication_year = str(arxiv[i]['published']).replace('\n', ' ')
                 paperInGeneral_md += f'''|{i + 1}|{title}|{publication_year}|\n'''
-        if 'PaperWithCode' in platforms:
+        if 'Paper with Code' in platforms:
             paperInGeneral_md += '''
 ## Paper with Code
 | ID| Paper Title | Publication Year |
@@ -72,6 +73,8 @@ def render_body(platforms, num_papers, num_papers_preview, query_input, show_pre
             st.markdown(f'''## Clusters Overview''')
             st.markdown(f'''Here we show the overview of the clusters, more specifically,''')
             st.markdown(f'''\n- the number of papers in each cluster\n- the number of keyphrases of each cluster''')
+            '''
+            plot using pyecharts
             bar = (
                 Bar()
                 .add_xaxis([f'Cluster {i + 1}' for i in range(len(clusters))])
@@ -80,6 +83,12 @@ def render_body(platforms, num_papers, num_papers_preview, query_input, show_pre
             )
             html = generate_html_pyecharts(bar, 'tmp.html')
             st_render.html(html, height=500, width=1000)
+            '''
+            st.bokeh_chart(build_bar_charts(
+                x_range=[f'Cluster {i + 1}' for i in range(len(clusters))],
+                y_names= ['Number of Papers', 'Number of Keyphrases'],
+                y_data=[[len(c) for c in clusters],[len(c.get_keyphrases()) for c in clusters]]
+            ))
 
 
 

widgets/charts.py

@@ -0,0 +1,80 @@
+from typing import List
+from bokeh.models import ColumnDataSource
+from bokeh.plotting import figure
+from bokeh.transform import dodge
+import numpy as np
+
+COLORS = [
+    '#FE2D01',
+    '#016CFE',
+    '#FEB101',
+    '#FE018B',
+    '#AAB7B8',
+    '#212F3D'
+]
+
+'''
+    clusters = ['Cluster 1', 'C 2', 'C 3', 'Plums', 'Grapes', 'Strawberries']
+    years = ['number of papers', 'number of keyphrases', ]
+
+    data = {'clusters': clusters,
+            f'{years[0]}': [2, 1, 4, 3, 2, 4],
+            f'{years[1]}': [5, 3, 3, 2, 4, 6],
+            }
+
+    source = ColumnDataSource(data=data)
+
+    p = figure(x_range=clusters, title="Fruit counts by year",
+               toolbar_location=None, tools="")
+
+    p.vbar(x=dodge('clusters', -0.25, range=p.x_range), top=f'{years[0]}', width=0.2, source=source,
+           color="#c9d9d3", legend_label="2015")
+
+    p.vbar(x=dodge('clusters', 0.0, range=p.x_range), top=f'{years[1]}', width=0.2, source=source,
+           color="#718dbf", legend_label="2016")
+
+
+    p.x_range.range_padding = 0.1
+    p.xgrid.grid_line_color = None
+    p.legend.location = "top_left"
+    p.legend.orientation = "horizontal"
+'''
+
+
+def build_bar_charts(x_range: List, y_names: List[str], y_data = List[List]):
+    valid_y = lambda x: len(x) == len(x_range)
+    if not (len(y_names) == len(y_data) and all(map(valid_y,y_data))):
+        raise RuntimeError('The data shapes are not aligned.')
+
+
+    if len(y_names) % 2 == 0:
+        offsets = [-0.125 - 0.25*(i-1) for i in range(len(y_names)//2,0,-1)]
+        offsets += [0.125 + 0.25*(i) for i in range(len(y_names)//2)]
+    else:
+        offsets = [-0.25 * i for i in range(len(y_names)//2,0,-1)]
+        offsets.append(0)
+        offsets += [0.25* (i+1) for i in range(len(y_names)//2)]
+
+    data = {
+        'x': x_range
+    }
+    for i,y in enumerate(y_data):
+        data[f'y{i}'] = y
+    source = ColumnDataSource(data)
+    p = figure(x_range=x_range,
+               tools = "box_zoom,save,reset",
+               height=500,
+               y_range=(0,np.max(y_data)+10)
+               )
+
+    for i,y in enumerate(y_data):
+        p.vbar(x=dodge('x', offsets[i], range=p.x_range), top=f'y{i}', width=0.2, source=source,
+               color=COLORS[i], legend_label=y_names[i])
+
+    p.x_range.range_padding = 0.1
+    p.xgrid.grid_line_color = None
+    p.legend.location = "top_left"
+    p.legend.orientation = "horizontal"
+
+    return p
+