初步结果

app.py

@@ -22,11 +22,13 @@ def main():
     # https://www.cnblogs.com/wang_yb/p/18502232
 
     page1 = st.Page("my_pages/page1.py", title="无监督聚类动态变化", icon="📊", default=True)
-    page2 = st.Page("my_pages/page2.py", title="第二个子应用", icon="📈")
-    page3 = st.Page("my_pages/page3.py", title="第三个子应用", icon="📉")
+    page2 = st.Page("my_pages/page2.py", title="超图神经网络训练过程", icon="📈")
+    # page3 = st.Page("my_pages/page3.py", title="超图构建分析-cora", icon="📉")
     # page3 = st.Page("dev/test.py", title="第三个子应用", icon="📉")
 
-    pg = st.navigation([page1, page2, page3])
+    pg = st.navigation([page1, page2, 
+                        # page3
+                        ])
     pg.run()
     # pass
 

my_pages/page1.py

@@ -10,10 +10,12 @@ from io import BytesIO
 import time
 from utils.data_processor import load_data, process_data, build_hyperedges
 from utils.visualizer import visualize_gmm, visualize_ratings
-from utils.hypergraph_drawer import draw_hypergraph, pyplot_fig_to_buffer
+
+from utils.streamlit_hypergraph import hypergraph_visualization_component
+
 
 def main():
-    st.title("NeurIPS 论文数据集高斯混合聚类分析")
+    st.title("NeurIPS 2024 Bench Paper 高斯混合聚类分析")
 
     # 自动播放
     slider_max = 10
@@ -34,7 +36,6 @@ def main():
         button_label = "开始拟合"
     st.button(button_label, on_click=toggle_play, key="play_button")
     # 播放速度
-    speed = st.slider("拟合速度", min_value=0.1, max_value=2.0, value=1.0, step=0.1, key="speed_slider")
     # speed = st.slider("播放速度", min_value=0.1, max_value=2.0, value=1.0, step=0.1, key="speed_slider")
     # 主页面布局
 
@@ -53,6 +54,7 @@ def main():
     # 使用 sidebar 控制参数
     with st.sidebar:
         st.header("控制面板")
+        speed = st.slider("拟合速度", min_value=0.1, max_value=2.0, value=1.0, step=0.1, key="speed_slider")
         draw_width = st.slider("绘图宽度", min_value=3, max_value=20, value=6, step=1, key="draw_width")
         draw_height = st.slider("绘图高度",min_value=3, max_value=20, value=6, step=1, key="draw_height")
 
@@ -74,7 +76,8 @@ def main():
         # Top K Clusters
         if display_option == "Top K Clusters":
             max_k = 5
-            top_k = st.slider("选择 K 值", min_value=1, max_value=max_k, value=3, step=1)
+            top_k = st.slider("选择 K 值", min_value=1, max_value=max_k, 
+                              value=1, step=1)
             top_p = None
         else:
             top_k = None
@@ -88,73 +91,16 @@ def main():
 
         # print(hyperedges)
 
+    show_hypergraph = st.checkbox("显示超图", value=True, key="show_hyperedges")
+    show_gaussian = st.checkbox("显示高斯分布", value=False, key="show_gaussian")
 
-    st.header("超图可视化")
-
-    draw_dual = st.toggle("对偶翻转", value=False, key="draw_dual")
-    euler_first = st.toggle("先看Euler Diagram", value=True, key="euler_first")
+    if show_hypergraph:
+        hypergraph_visualization_component(hypergraph, draw_width, draw_height)
     
-    def euler_visualization(hypergraph):
-        st.subheader("Euler Diagram 可视化")
-        col1, col2, col3, col4 = st.columns(4)
-        with col1:
-            fill_edges  = st.toggle("填充超边", value=True, key="fill_edges")
-        with col2:
-            toplexes    = st.toggle("只保留未被包含的边（toplexes）", value=False, key="toplexes")
-        with col3:
-            with_edge_labels = st.toggle("不看边名", value=False, key="with_edge_labels")
-            
-
-        col1, col2 = st.columns(2)
-        captions = ["论文聚类超图", "论文聚类对偶超图"]
-
-        if 'poses' not in st.session_state:
-            st.session_state.poses = [None, None]
-        if draw_dual:
-            hypergraph = hypergraph.dual()
-            captions = ["论文聚类对偶超图", "论文聚类超图"]
-            # st.session_state.poses = [st.session_state.poses[1], st.session_state.poses[0]]
-
-        with col1:
-            hypergraph_image, st.session_state.poses[0] = draw_hypergraph(hypergraph, draw_dual=False, 
-                                            fill_edges=fill_edges, toplexes=toplexes,
-                                            with_edge_labels=with_edge_labels
-                                            # pos=st.session_state.poses[0]
-                                            ,draw_width=draw_width, draw_height=draw_height
-                                            )
-            st.image(hypergraph_image, caption = captions[0], use_container_width=True)
-
-        with col2:
-            hypergraph_image, st.session_state.poses[1] = draw_hypergraph(hypergraph, draw_dual=True, 
-                                            fill_edges=fill_edges, toplexes=toplexes,
-                                            with_edge_labels=with_edge_labels
-                                            # pos=st.session_state.poses[1]
-                                            ,draw_width=draw_width, draw_height=draw_height
-                                            )
-            st.image(hypergraph_image, caption =captions[1], use_container_width=True)
-
-    def new_visualization(hypergraph):
-        st.subheader("新型可视化")
-        col1, col2 = st.columns(2)
-        with col1:
-            fig, ax = plt.subplots(figsize=(draw_width, draw_height))
-            hnx.draw_bipartite_using_euler(hypergraph)
-            st.image(pyplot_fig_to_buffer(plt.gcf()), caption="双列二分图可视化", use_container_width=True)
-        with col2:
-            fig, ax = plt.subplots(figsize=(draw_width, draw_height))
-            hnx.draw_incidence_upset(hypergraph)
-            st.image(pyplot_fig_to_buffer(plt.gcf()), caption="Incidence/UpSet 可视化", use_container_width=True)
-
-    if euler_first:
-        euler_visualization(hypergraph)
-        new_visualization(hypergraph)
-    else:
-        new_visualization(hypergraph)
-        euler_visualization(hypergraph)
-
-    st.header("高斯混合分布聚类结果")
-    fig_gmm = visualize_gmm(sampled_df, iteration)
-    st.plotly_chart(fig_gmm, use_container_width=True)
+    if show_gaussian:
+        st.header("高斯混合分布聚类结果")
+        fig_gmm = visualize_gmm(sampled_df, iteration)
+        st.plotly_chart(fig_gmm, use_container_width=True)
 
     # 显示采样论文的详细信息
     st.header("采样论文详细信息")

my_pages/page2.py

@@ -1,7 +1,139 @@
+from typing import Dict, List
 import streamlit as st
+import pandas as pd
+import numpy as np
+import plotly.express as px
+import hypernetx as hnx
+import matplotlib.pyplot as plt
+from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
+from io import BytesIO
+import time
+import json
+from utils.data_processor import process_data, build_hyperedges
+from utils.visualizer import visualize_gmm, visualize_ratings
+from utils.streamlit_hypergraph import hypergraph_visualization_component
+
+def load_json_data(file_path: str):
+    """从JSON文件中加载数据"""
+    with open(file_path, 'r') as f:
+        data = json.load(f)
+    return data
+
+
+
 
 def main():
-    st.title("第二个子应用")
-    st.write("这里是第二个子应用的内容。")
+    with st.sidebar:
+        st.header("控制面板")
+        dataset_selection = st.selectbox(
+            "选择可视化的数据集",
+            ["NeurIPS 2024 Bench", "Cora Co-Author"]
+        )
+    st.title(f"{dataset_selection} Author-Paper 超图可视化分析")
+
+    # 自动播放状态
+    slider_max = 200
+    if 'play_state' not in st.session_state:
+        st.session_state.play_state = False
+    if 'iteration' not in st.session_state:
+        st.session_state.iteration = 10
+
+    # 定义回调函数来切换播放状态
+    def toggle_play():
+        if not st.session_state.play_state and st.session_state.iteration == slider_max:
+            st.session_state.iteration = 10  # 重置迭代次数
+        st.session_state.play_state = not st.session_state.play_state
+
+    # 创建播放/暂停按钮
+    if st.session_state.play_state:
+        button_label = "暂停"
+    else:
+        button_label = "开始拟合"
+    st.button(button_label, on_click=toggle_play, key="play_button")
+
+    # 显示迭代次数滑条
+    iteration = st.slider("迭代步骤", min_value=10, max_value=slider_max, 
+                          value=st.session_state.iteration, step=10, 
+                          key="iteration_slider")
+
+    # 从JSON文件加载数据
+    # 假设JSON文件路径为"authors_papers.json"
+    if "hyper_edges" not in st.session_state or "labels_history" not in st.session_state:
+        if dataset_selection == "NeurIPS 2024 Bench":
+            json_hyper_edges = load_json_data("hyper_edges_nips.json")
+            json_labels_history:list = load_json_data("labels_history_nips.json")
+        elif dataset_selection == "Cora Co-Author":
+            json_hyper_edges = load_json_data("hyper_edges.json")
+            json_labels_history:list = load_json_data("labels_history.json")
+        else:
+            st.error("未知数据集")
+            raise ValueError("Unknown dataset selected")
+
+        st.session_state.hyper_edges = json_hyper_edges
+        
+        st.session_state.labels_history = {
+            item['epoch']:item["labels"]
+            for item in json_labels_history
+        }
+        
+    
+    # 使用 sidebar 控制参数
+    with st.sidebar:
+        speed = st.slider("拟合速度", min_value=0.1, max_value=2.0, value=1.0, step=0.1, key="speed_slider")
+        draw_width = st.slider("绘图宽度", min_value=3, max_value=20, value=6, step=1, key="draw_width")
+        draw_height = st.slider("绘图高度", min_value=3, max_value=20, value=6, step=1, key="draw_height")
+
+        max_samples = len(st.session_state.hyper_edges)
+        num_samples = st.slider("选择采样作者数量", min_value=1, 
+                                max_value=min(100, max_samples), value=min(10, max_samples), step=1)
+        
+        # 采样部分作者
+        sampled_authors = np.random.choice(st.session_state.hyper_edges, size=num_samples, replace=False)
+        
+        show_labels = st.checkbox("展示分类结果", value=True, key="show_labels")
+        
+        def paper_list_to_types(iteration:int, paper_list:List[int]) -> List[str]:
+            """将论文列表转换为类型列表"""
+            labels:List[int] = st.session_state.labels_history[str(iteration)]
+            return [labels[paper] for paper in paper_list]
+        
+        
+        sampled_data = {list(author_dict.keys()
+        )[0]: list(author_dict.values()
+        )[0] if not show_labels else paper_list_to_types(iteration, list(author_dict.values())[0])
+        for author_dict in sampled_authors}
+        
+        # 构建超边
+        hyperedges = sampled_data
+
+
+        # 修改 类别信息
+
+
+
+
+        hypergraph = hnx.Hypergraph(hyperedges)
+
+    show_hypergraph = st.checkbox("显示超图", value=True, key="show_hyperedges")
+
+    if show_hypergraph:
+        hypergraph_visualization_component(hypergraph, draw_width, draw_height)
+    
+    # 显示采样作者的详细信息
+    st.header("采样作者详细信息")
+    authors_df = pd.DataFrame([(author, len(papers)) for author, papers in sampled_data.items()], 
+                             columns=["作者", "论文数量"])
+    st.dataframe(authors_df)
+
+    # 自动播放功能
+    if st.session_state.play_state:
+        with st.spinner("正在播放..."):
+            if st.session_state.iteration < slider_max:
+                st.session_state.iteration += 10
+                st.write(f"当前迭代次数: {st.session_state.iteration}")
+                time.sleep(1/speed)  # 根据速度调整等待时间
+                st.rerun()
+            else:
+                st.session_state.play_state = False
 
 main()

my_pages/page3.py

@@ -1,7 +1,95 @@
+from typing import Dict, List
 import streamlit as st
+import pandas as pd
+import numpy as np
+import plotly.express as px
+import hypernetx as hnx
+import matplotlib.pyplot as plt
+from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
+from io import BytesIO
+import time
+import json
+from utils.data_processor import process_data, build_hyperedges
+from utils.visualizer import visualize_gmm, visualize_ratings
+from utils.streamlit_hypergraph import hypergraph_visualization_component
+
+def load_json_data(file_path: str) -> Dict[str, List[int]]:
+    """从JSON文件中加载数据"""
+    with open(file_path, 'r') as f:
+        data = json.load(f)
+    return data
 
 def main():
-    st.title("第三个子应用")
-    st.write("这里是第三个子应用的内容。")
+    st.title("CORA Author-Paper 超图可视化分析")
+
+    # 自动播放状态
+    slider_max = 10
+    if 'play_state' not in st.session_state:
+        st.session_state.play_state = False
+    if 'iteration' not in st.session_state:
+        st.session_state.iteration = 0
+
+    # 定义回调函数来切换播放状态
+    def toggle_play():
+        if not st.session_state.play_state and st.session_state.iteration == slider_max:
+            st.session_state.iteration = 0  # 重置迭代次数
+        st.session_state.play_state = not st.session_state.play_state
+
+    # 创建播放/暂停按钮
+    if st.session_state.play_state:
+        button_label = "暂停"
+    else:
+        button_label = "开始拟合"
+    st.button(button_label, on_click=toggle_play, key="play_button")
+
+    # 显示迭代次数滑条
+    iteration = st.slider("迭代步骤", min_value=1, max_value=slider_max, 
+                          value=st.session_state.iteration, step=1, 
+                          key="iteration_slider")
+
+    # 从JSON文件加载数据
+    # 假设JSON文件路径为"authors_papers.json"
+    data = load_json_data("hyper_edges.json")
+    
+    # 使用 sidebar 控制参数
+    with st.sidebar:
+        st.header("控制面板")
+        speed = st.slider("拟合速度", min_value=0.1, max_value=2.0, value=1.0, step=0.1, key="speed_slider")
+        draw_width = st.slider("绘图宽度", min_value=3, max_value=20, value=6, step=1, key="draw_width")
+        draw_height = st.slider("绘图高度", min_value=3, max_value=20, value=6, step=1, key="draw_height")
+
+        max_samples = len(data)
+        num_samples = st.slider("选择采样作者数量", min_value=1, 
+                                max_value=min(100, max_samples), value=min(10, max_samples), step=1)
+        
+        # 采样部分作者
+        sampled_authors = np.random.choice(data, size=num_samples, replace=False)
+        sampled_data = {list(author_dict.keys())[0]: list(author_dict.values())[0] for author_dict in sampled_authors}
+        
+        # 构建超边
+        hyperedges = sampled_data
+        hypergraph = hnx.Hypergraph(hyperedges)
+
+    show_hypergraph = st.checkbox("显示超图", value=True, key="show_hyperedges")
+
+    if show_hypergraph:
+        hypergraph_visualization_component(hypergraph, draw_width, draw_height)
+    
+    # 显示采样作者的详细信息
+    st.header("采样作者详细信息")
+    authors_df = pd.DataFrame([(author, len(papers)) for author, papers in sampled_data.items()], 
+                             columns=["作者", "论文数量"])
+    st.dataframe(authors_df)
+
+    # 自动播放功能
+    if st.session_state.play_state:
+        with st.spinner("正在播放..."):
+            if st.session_state.iteration < slider_max:
+                st.session_state.iteration += 1
+                st.write(f"当前迭代次数: {st.session_state.iteration}")
+                time.sleep(1/speed)  # 根据速度调整等待时间
+                st.rerun()
+            else:
+                st.session_state.play_state = False
 
 main()

utils/streamlit_hypergraph.py

@@ -0,0 +1,76 @@
+from matplotlib import pyplot as plt
+import streamlit as st
+import hypernetx as hnx
+from utils.hypergraph_drawer import draw_hypergraph, pyplot_fig_to_buffer
+
+def hypergraph_visualization_component(hypergraph:hnx.Hypergraph, 
+                         draw_width:int, 
+                         draw_height:int):
+    """
+    Visualize a hypergraph using Streamlit.
+    
+    Parameters:
+    hypergraph (dict): A dictionary representing the hypergraph, where keys are nodes and values are lists of connected nodes.
+    """
+    st.header("超图可视化")
+
+    draw_dual = st.toggle("对偶翻转", value=False, key="draw_dual")
+    euler_first = st.toggle("先看Euler Diagram", value=True, key="euler_first")
+    if draw_dual:
+        hypergraph = hypergraph.dual()
+    def euler_visualization(hypergraph):
+        st.subheader("Euler Diagram 可视化")
+        col1, col2, col3, col4 = st.columns(4)
+        with col1:
+            fill_edges  = st.toggle("填充超边", value=True, key="fill_edges")
+        with col2:
+            toplexes    = st.toggle("只保留未被包含的边（toplexes）", value=False, key="toplexes")
+        with col3:
+            with_edge_labels = st.toggle("不看边名", value=False, key="with_edge_labels")
+            
+
+        col1, col2 = st.columns(2)
+        captions = ["超图", "对偶超图"]
+
+        if 'poses' not in st.session_state:
+            st.session_state.poses = [None, None]
+        if draw_dual:
+            captions = ["对偶超图", "超图"]
+            # st.session_state.poses = [st.session_state.poses[1], st.session_state.poses[0]]
+
+        with col1:
+            hypergraph_image, st.session_state.poses[0] = draw_hypergraph(hypergraph, draw_dual=False, 
+                                            fill_edges=fill_edges, toplexes=toplexes,
+                                            with_edge_labels=with_edge_labels
+                                            # pos=st.session_state.poses[0]
+                                            ,draw_width=draw_width, draw_height=draw_height
+                                            )
+            st.image(hypergraph_image, caption = captions[0], use_container_width=True)
+
+        with col2:
+            hypergraph_image, st.session_state.poses[1] = draw_hypergraph(hypergraph, draw_dual=True, 
+                                            fill_edges=fill_edges, toplexes=toplexes,
+                                            with_edge_labels=with_edge_labels
+                                            # pos=st.session_state.poses[1]
+                                            ,draw_width=draw_width, draw_height=draw_height
+                                            )
+            st.image(hypergraph_image, caption =captions[1], use_container_width=True)
+
+    def new_visualization(hypergraph):
+        st.subheader("新型可视化")
+        col1, col2 = st.columns(2)
+        with col1:
+            fig, ax = plt.subplots(figsize=(draw_width, draw_height))
+            hnx.draw_bipartite_using_euler(hypergraph)
+            st.image(pyplot_fig_to_buffer(plt.gcf()), caption="双列二分图可视化", use_container_width=True)
+        with col2:
+            fig, ax = plt.subplots(figsize=(draw_width, draw_height))
+            hnx.draw_incidence_upset(hypergraph)
+            st.image(pyplot_fig_to_buffer(plt.gcf()), caption="Incidence/UpSet 可视化", use_container_width=True)
+
+    if euler_first:
+        euler_visualization(hypergraph)
+        new_visualization(hypergraph)
+    else:
+        new_visualization(hypergraph)
+        euler_visualization(hypergraph)