Update app.py
Browse files
app.py
CHANGED
|
@@ -1,117 +1,117 @@
|
|
| 1 |
-
import gradio as gr
|
| 2 |
-
from utils.embedding import ChatEmbedder
|
| 3 |
-
from utils.retrieval import ChatQABot
|
| 4 |
-
import json
|
| 5 |
-
|
| 6 |
-
class ChatAnalyzerApp:
|
| 7 |
-
def __init__(self):
|
| 8 |
-
self.embedder = ChatEmbedder()
|
| 9 |
-
self.qa_bot = ChatQABot()
|
| 10 |
-
self.setup_data()
|
| 11 |
-
|
| 12 |
-
def setup_data(self):
|
| 13 |
-
"""初始化数据和索引"""
|
| 14 |
-
try:
|
| 15 |
-
count = self.embedder.load_chat_data('chat_log.json')
|
| 16 |
-
index_size = self.embedder.build_index()
|
| 17 |
-
print(f"✓ 加载了 {count} 条聊天记录,构建了 {index_size} 维索引")
|
| 18 |
-
return True
|
| 19 |
-
except Exception as e:
|
| 20 |
-
print(f"× 数据加载失败: {e}")
|
| 21 |
-
return False
|
| 22 |
-
|
| 23 |
-
def query_chat(self, question: str, top_k: int = 3):
|
| 24 |
-
"""处理查询"""
|
| 25 |
-
# 1. 检索相关文档
|
| 26 |
-
results = self.embedder.search(question, top_k)
|
| 27 |
-
|
| 28 |
-
if not results:
|
| 29 |
-
return "未找到相关聊天记录。", ""
|
| 30 |
-
|
| 31 |
-
# 2. 组合检索到的上下文
|
| 32 |
-
context = "\n".join([f"{i+1}. {r['content']} (相关度: {r['score']:.3f})"
|
| 33 |
-
for i, r in enumerate(results)])
|
| 34 |
-
|
| 35 |
-
# 3. 生成答案
|
| 36 |
-
answer = self.qa_bot.generate_answer(question, context)
|
| 37 |
-
|
| 38 |
-
return answer, context
|
| 39 |
-
|
| 40 |
-
def main():
|
| 41 |
-
app = ChatAnalyzerApp()
|
| 42 |
-
|
| 43 |
-
if not app.setup_data():
|
| 44 |
-
return
|
| 45 |
-
|
| 46 |
-
# 创建Gradio界面
|
| 47 |
-
with gr.Blocks(title="群聊记录分析助手", theme=gr.themes.Soft()) as demo:
|
| 48 |
-
gr.Markdown("# 📱 群聊记录分析与问答系统")
|
| 49 |
-
gr.Markdown("基于RAG技术,智能分析群聊记录并回答问题")
|
| 50 |
-
|
| 51 |
-
with gr.Row():
|
| 52 |
-
with gr.Column(scale=2):
|
| 53 |
-
question_input = gr.Textbox(
|
| 54 |
-
label="输入您的问题",
|
| 55 |
-
placeholder="例如:关于新版UI设计,最终的结论是什么?",
|
| 56 |
-
lines=2
|
| 57 |
-
)
|
| 58 |
-
|
| 59 |
-
top_k_slider = gr.Slider(
|
| 60 |
-
minimum=1,
|
| 61 |
-
maximum=5,
|
| 62 |
-
value=3,
|
| 63 |
-
step=1,
|
| 64 |
-
label="检索结果数量"
|
| 65 |
-
)
|
| 66 |
-
|
| 67 |
-
submit_btn = gr.Button("🔍 搜索", variant="primary")
|
| 68 |
-
|
| 69 |
-
with gr.Column(scale=3):
|
| 70 |
-
answer_output = gr.Textbox(
|
| 71 |
-
label="答案",
|
| 72 |
-
lines=4,
|
| 73 |
-
interactive=False
|
| 74 |
-
)
|
| 75 |
-
|
| 76 |
-
context_output = gr.Textbox(
|
| 77 |
-
label="检索到的相关聊天记录",
|
| 78 |
-
lines=6,
|
| 79 |
-
interactive=False
|
| 80 |
-
)
|
| 81 |
-
|
| 82 |
-
# 示例问题
|
| 83 |
-
examples = gr.Examples(
|
| 84 |
-
examples=[
|
| 85 |
-
"关于新版UI设计,最终的结论是什么?",
|
| 86 |
-
"谁负责联系市场部?",
|
| 87 |
-
"市场部的联系方式是什么?",
|
| 88 |
-
"讨论过哪些设计方案?"
|
| 89 |
-
],
|
| 90 |
-
inputs=[question_input]
|
| 91 |
-
)
|
| 92 |
-
|
| 93 |
-
gr.Markdown("### 功能说明")
|
| 94 |
-
gr.Markdown("""
|
| 95 |
-
1. 系统基于模拟的飞书/企微群聊记录构建知识库
|
| 96 |
-
2. 使用中文嵌入模型检索相关聊天片段
|
| 97 |
-
3. 使用大语言模型生成准确答案
|
| 98 |
-
4. 支持调整检索结果数量以优化答案质量
|
| 99 |
-
""")
|
| 100 |
-
|
| 101 |
-
# 绑定事件
|
| 102 |
-
submit_btn.click(
|
| 103 |
-
fn=app.query_chat,
|
| 104 |
-
inputs=[question_input, top_k_slider],
|
| 105 |
-
outputs=[answer_output, context_output]
|
| 106 |
-
)
|
| 107 |
-
|
| 108 |
-
question_input.submit(
|
| 109 |
-
fn=app.query_chat,
|
| 110 |
-
inputs=[question_input, top_k_slider],
|
| 111 |
-
outputs=[answer_output, context_output]
|
| 112 |
-
)
|
| 113 |
-
|
| 114 |
-
demo.launch(
|
| 115 |
-
|
| 116 |
-
if __name__ == "__main__":
|
| 117 |
main()
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from utils.embedding import ChatEmbedder
|
| 3 |
+
from utils.retrieval import ChatQABot
|
| 4 |
+
import json
|
| 5 |
+
|
| 6 |
+
class ChatAnalyzerApp:
|
| 7 |
+
def __init__(self):
|
| 8 |
+
self.embedder = ChatEmbedder()
|
| 9 |
+
self.qa_bot = ChatQABot()
|
| 10 |
+
self.setup_data()
|
| 11 |
+
|
| 12 |
+
def setup_data(self):
|
| 13 |
+
"""初始化数据和索引"""
|
| 14 |
+
try:
|
| 15 |
+
count = self.embedder.load_chat_data('chat_log.json')
|
| 16 |
+
index_size = self.embedder.build_index()
|
| 17 |
+
print(f"✓ 加载了 {count} 条聊天记录,构建了 {index_size} 维索引")
|
| 18 |
+
return True
|
| 19 |
+
except Exception as e:
|
| 20 |
+
print(f"× 数据加载失败: {e}")
|
| 21 |
+
return False
|
| 22 |
+
|
| 23 |
+
def query_chat(self, question: str, top_k: int = 3):
|
| 24 |
+
"""处理查询"""
|
| 25 |
+
# 1. 检索相关文档
|
| 26 |
+
results = self.embedder.search(question, top_k)
|
| 27 |
+
|
| 28 |
+
if not results:
|
| 29 |
+
return "未找到相关聊天记录。", ""
|
| 30 |
+
|
| 31 |
+
# 2. 组合检索到的上下文
|
| 32 |
+
context = "\n".join([f"{i+1}. {r['content']} (相关度: {r['score']:.3f})"
|
| 33 |
+
for i, r in enumerate(results)])
|
| 34 |
+
|
| 35 |
+
# 3. 生成答案
|
| 36 |
+
answer = self.qa_bot.generate_answer(question, context)
|
| 37 |
+
|
| 38 |
+
return answer, context
|
| 39 |
+
|
| 40 |
+
def main():
|
| 41 |
+
app = ChatAnalyzerApp()
|
| 42 |
+
|
| 43 |
+
if not app.setup_data():
|
| 44 |
+
return
|
| 45 |
+
|
| 46 |
+
# 创建Gradio界面
|
| 47 |
+
with gr.Blocks(title="群聊记录分析助手", theme=gr.themes.Soft()) as demo:
|
| 48 |
+
gr.Markdown("# 📱 群聊记录分析与问答系统")
|
| 49 |
+
gr.Markdown("基于RAG技术,智能分析群聊记录并回答问题")
|
| 50 |
+
|
| 51 |
+
with gr.Row():
|
| 52 |
+
with gr.Column(scale=2):
|
| 53 |
+
question_input = gr.Textbox(
|
| 54 |
+
label="输入您的问题",
|
| 55 |
+
placeholder="例如:关于新版UI设计,最终的结论是什么?",
|
| 56 |
+
lines=2
|
| 57 |
+
)
|
| 58 |
+
|
| 59 |
+
top_k_slider = gr.Slider(
|
| 60 |
+
minimum=1,
|
| 61 |
+
maximum=5,
|
| 62 |
+
value=3,
|
| 63 |
+
step=1,
|
| 64 |
+
label="检索结果数量"
|
| 65 |
+
)
|
| 66 |
+
|
| 67 |
+
submit_btn = gr.Button("🔍 搜索", variant="primary")
|
| 68 |
+
|
| 69 |
+
with gr.Column(scale=3):
|
| 70 |
+
answer_output = gr.Textbox(
|
| 71 |
+
label="答案",
|
| 72 |
+
lines=4,
|
| 73 |
+
interactive=False
|
| 74 |
+
)
|
| 75 |
+
|
| 76 |
+
context_output = gr.Textbox(
|
| 77 |
+
label="检索到的相关聊天记录",
|
| 78 |
+
lines=6,
|
| 79 |
+
interactive=False
|
| 80 |
+
)
|
| 81 |
+
|
| 82 |
+
# 示例问题
|
| 83 |
+
examples = gr.Examples(
|
| 84 |
+
examples=[
|
| 85 |
+
"关于新版UI设计,最终的结论是什么?",
|
| 86 |
+
"谁负责联系市场部?",
|
| 87 |
+
"市场部的联系方式是什么?",
|
| 88 |
+
"讨论过哪些设计方案?"
|
| 89 |
+
],
|
| 90 |
+
inputs=[question_input]
|
| 91 |
+
)
|
| 92 |
+
|
| 93 |
+
gr.Markdown("### 功能说明")
|
| 94 |
+
gr.Markdown("""
|
| 95 |
+
1. 系统基于模拟的飞书/企微群聊记录构建知识库
|
| 96 |
+
2. 使用中文嵌入模型检索相关聊天片段
|
| 97 |
+
3. 使用大语言模型生成准确答案
|
| 98 |
+
4. 支持调整检索结果数量以优化答案质量
|
| 99 |
+
""")
|
| 100 |
+
|
| 101 |
+
# 绑定事件
|
| 102 |
+
submit_btn.click(
|
| 103 |
+
fn=app.query_chat,
|
| 104 |
+
inputs=[question_input, top_k_slider],
|
| 105 |
+
outputs=[answer_output, context_output]
|
| 106 |
+
)
|
| 107 |
+
|
| 108 |
+
question_input.submit(
|
| 109 |
+
fn=app.query_chat,
|
| 110 |
+
inputs=[question_input, top_k_slider],
|
| 111 |
+
outputs=[answer_output, context_output]
|
| 112 |
+
)
|
| 113 |
+
|
| 114 |
+
demo.launch()
|
| 115 |
+
|
| 116 |
+
if __name__ == "__main__":
|
| 117 |
main()
|