DocUA commited on
Commit
212d350
·
1 Parent(s): 80fd9a9

Оновлено типи аналізу за замовчуванням у Gradio інтерфейсі на "Content Summary" та "Gemini 2.0 Flash". Внесено зміни до системних повідомлень для покращення контексту аналізу, щоб відображати мову оригінального документа. Видалено застаріле повідомлення про виправлення у відображенні результатів.

Browse files
Files changed (3) hide show
  1. app_interface.py +2 -16
  2. app_logic.py +2 -2
  3. llm/gemini_connector.py +4 -4
app_interface.py CHANGED
@@ -552,7 +552,7 @@ class MarkItDownTestingApp:
552
  ("Content Summary - Тематичний аналіз та ключові інсайти", "content_summary"),
553
  ("Extraction Quality - Оцінка збереження даних", "extraction_quality"),
554
  ],
555
- value="quality_analysis",
556
  label="Analysis Type",
557
  interactive=True,
558
  )
@@ -565,7 +565,7 @@ class MarkItDownTestingApp:
565
  ("Gemini 1.5 Pro (Legacy)", GeminiModel.LEGACY_PRO.value),
566
  ("Gemini 1.5 Flash (Legacy)", GeminiModel.LEGACY_FLASH.value),
567
  ],
568
- value=GeminiModel.PRO.value,
569
  label="AI Model Preference",
570
  interactive=True,
571
  )
@@ -586,20 +586,6 @@ class MarkItDownTestingApp:
586
  with gr.Column(scale=2):
587
  gr.Markdown("### 📊 Processing Results")
588
 
589
- # 🚨 ДОДАНО ВАЖЛИВЕ ПОВІДОМЛЕННЯ ПРО ВИПРАВЛЕННЯ
590
- gr.HTML("""
591
- <div style="background: #d1ecf1; border: 1px solid #bee5eb; padding: 15px; border-radius: 8px; margin-bottom: 20px;">
592
- <h4 style="margin: 0 0 10px 0; color: #0c5460;">🔧 Architectural Fix Applied</h4>
593
- <p style="margin: 0; color: #0c5460;"><strong>Fixed Issue:</strong> Different analysis types now show different results in Markdown Output!</p>
594
- <ul style="margin: 10px 0 0 20px; color: #0c5460;">
595
- <li><strong>Quality Analysis:</strong> Shows detailed quality metrics and AI feedback</li>
596
- <li><strong>Structure Review:</strong> Shows document structure analysis and organization</li>
597
- <li><strong>Content Summary:</strong> Shows thematic analysis and key insights</li>
598
- <li><strong>Extraction Quality:</strong> Shows data preservation assessment</li>
599
- </ul>
600
- </div>
601
- """)
602
-
603
  status_display = gr.HTML()
604
 
605
  with gr.Tabs():
 
552
  ("Content Summary - Тематичний аналіз та ключові інсайти", "content_summary"),
553
  ("Extraction Quality - Оцінка збереження даних", "extraction_quality"),
554
  ],
555
+ value="content_summary",
556
  label="Analysis Type",
557
  interactive=True,
558
  )
 
565
  ("Gemini 1.5 Pro (Legacy)", GeminiModel.LEGACY_PRO.value),
566
  ("Gemini 1.5 Flash (Legacy)", GeminiModel.LEGACY_FLASH.value),
567
  ],
568
+ value=GeminiModel.FLASH.value,
569
  label="AI Model Preference",
570
  interactive=True,
571
  )
 
586
  with gr.Column(scale=2):
587
  gr.Markdown("### 📊 Processing Results")
588
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
589
  status_display = gr.HTML()
590
 
591
  with gr.Tabs():
app_logic.py CHANGED
@@ -38,8 +38,8 @@ class ProcessingRequest:
38
  file_content: bytes
39
  file_metadata: JSONDict
40
  gemini_api_key: Optional[str] = None
41
- analysis_type: str = AnalysisType.QUALITY_ANALYSIS.value
42
- model_preference: str = GeminiModel.PRO.value
43
  use_llm: bool = False
44
  enable_plugins: bool = False
45
  azure_endpoint: Optional[str] = None
 
38
  file_content: bytes
39
  file_metadata: JSONDict
40
  gemini_api_key: Optional[str] = None
41
+ analysis_type: str = AnalysisType.CONTENT_SUMMARY.value
42
+ model_preference: str = GeminiModel.FLASH.value
43
  use_llm: bool = False
44
  enable_plugins: bool = False
45
  azure_endpoint: Optional[str] = None
llm/gemini_connector.py CHANGED
@@ -155,7 +155,7 @@ class GeminiAnalysisEngine:
155
  ANALYSIS_PROMPTS = {
156
  AnalysisType.QUALITY_ANALYSIS: {
157
  "system": """You are an expert document conversion analyst specializing in evaluating
158
- the quality of document-to-Markdown conversions.""",
159
  "template": """
160
  Analyze the quality of this Markdown conversion from a document.
161
 
@@ -187,7 +187,7 @@ class GeminiAnalysisEngine:
187
 
188
  AnalysisType.STRUCTURE_REVIEW: {
189
  "system": """You are a document structure specialist analyzing Markdown
190
- document organization and hierarchy.""",
191
  "template": """
192
  Conduct a comprehensive structural analysis of this Markdown document.
193
 
@@ -219,7 +219,7 @@ class GeminiAnalysisEngine:
219
 
220
  AnalysisType.CONTENT_SUMMARY: {
221
  "system": """You are a content analysis expert specializing in document
222
- summarization and thematic analysis.""",
223
  "template": """
224
  Create a comprehensive content summary and thematic analysis of this document.
225
 
@@ -251,7 +251,7 @@ class GeminiAnalysisEngine:
251
 
252
  AnalysisType.EXTRACTION_QUALITY: {
253
  "system": """You are a data extraction quality specialist evaluating how well
254
- information was preserved during document conversion.""",
255
  "template": """
256
  Evaluate the extraction quality and information preservation in this converted document.
257
 
 
155
  ANALYSIS_PROMPTS = {
156
  AnalysisType.QUALITY_ANALYSIS: {
157
  "system": """You are an expert document conversion analyst specializing in evaluating
158
+ the quality of document-to-Markdown conversions (use the language of the original document).""",
159
  "template": """
160
  Analyze the quality of this Markdown conversion from a document.
161
 
 
187
 
188
  AnalysisType.STRUCTURE_REVIEW: {
189
  "system": """You are a document structure specialist analyzing Markdown
190
+ document organization and hierarchy (use the language of the original document).""",
191
  "template": """
192
  Conduct a comprehensive structural analysis of this Markdown document.
193
 
 
219
 
220
  AnalysisType.CONTENT_SUMMARY: {
221
  "system": """You are a content analysis expert specializing in document
222
+ summarization and thematic analysis (use the language of the original document).""",
223
  "template": """
224
  Create a comprehensive content summary and thematic analysis of this document.
225
 
 
251
 
252
  AnalysisType.EXTRACTION_QUALITY: {
253
  "system": """You are a data extraction quality specialist evaluating how well
254
+ information was preserved during document conversion (use the language of the original document).""",
255
  "template": """
256
  Evaluate the extraction quality and information preservation in this converted document.
257