wehub-resource-sync
262 lines
13 KiB
JSON
262 lines
13 KiB
JSON
{
|
|
"extraction_configs": {
|
|
"scientific_report": {
|
|
"document_name": "Research findings document",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"title": "Title of the report",
|
|
"authors": "List of authors",
|
|
"publication_date": "Date of publication",
|
|
"abstract_summary": "Summary of the abstract or executive summary",
|
|
"keywords": "Key topics or keywords mentioned"
|
|
}
|
|
},
|
|
"resume": {
|
|
"document_name": "CV or work history",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"candidate_name": "Full name of the candidate",
|
|
"email": "Email address",
|
|
"phone": "Phone number",
|
|
"education": "Highest degree and university",
|
|
"skills": "List of technical or professional skills",
|
|
"last_job_title": "Most recent job title"
|
|
}
|
|
},
|
|
"memo": {
|
|
"document_name": "Internal office brief",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"sender": "Name or department sending the memo",
|
|
"recipient": "Name or department receiving the memo",
|
|
"date": "Date of the memo",
|
|
"subject": "Subject line",
|
|
"action_items": "List of required actions or requests"
|
|
}
|
|
},
|
|
"file_folder": {
|
|
"document_name": "Folder cover or label",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"label_text": "Text written on the folder tab or cover",
|
|
"department": "Department or category associated with the folder",
|
|
"year": "Year associated with the folder contents",
|
|
"color": "Visual color of the folder if distinguishable"
|
|
}
|
|
},
|
|
"specification": {
|
|
"document_name": "Design requirements",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"project_name": "Name of the project or product",
|
|
"spec_id": "Specification ID or document number",
|
|
"version": "Version number",
|
|
"requirements_summary": "Summary of main technical requirements",
|
|
"status": "Status (e.g., Draft, Approved, Deprecated)"
|
|
}
|
|
},
|
|
"news_article": {
|
|
"document_name": "News report",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"headline": "Main headline of the article",
|
|
"publication_source": "Name of the newspaper or website",
|
|
"date": "Date of publication",
|
|
"author": "Name of the journalist or author",
|
|
"topic": "Main topic or category (e.g., Politics, Sports, Tech)"
|
|
}
|
|
},
|
|
"letter": {
|
|
"document_name": "Written correspondence",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"sender_name": "Name of the sender",
|
|
"sender_address": "Address of the sender",
|
|
"recipient_name": "Name of the recipient",
|
|
"recipient_address": "Address of the recipient",
|
|
"date": "Date of the letter",
|
|
"subject": "Subject or reference line"
|
|
}
|
|
},
|
|
"form": {
|
|
"document_name": "Fill-in-the-blank document",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"form_title": "Title or header of the form",
|
|
"form_id": "Form ID number",
|
|
"person_name": "Name of the person filling the form",
|
|
"submission_date": "Date the form was filled or signed",
|
|
"is_signed": "Boolean indicating if a signature is present"
|
|
}
|
|
},
|
|
"budget": {
|
|
"document_name": "Financial plan",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"period_start": "Start date of the budget period",
|
|
"period_end": "End date of the budget period",
|
|
"total_revenue": "Total projected revenue",
|
|
"total_expenses": "Total projected expenses",
|
|
"department": "Department or project the budget belongs to"
|
|
}
|
|
},
|
|
"handwritten": {
|
|
"document_name": "Non-typed text",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"transcription": "Full text transcription of the handwriting",
|
|
"language": "Language of the text",
|
|
"sentiment": "Sentiment of the content (Positive, Negative, Neutral)",
|
|
"writer_name": "Name of the writer if identifiable"
|
|
}
|
|
},
|
|
"email": {
|
|
"document_name": "Digital message",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"from": "Sender email address",
|
|
"to": "Recipient email address",
|
|
"cc": "CC email addresses",
|
|
"date": "Date and time sent",
|
|
"subject": "Subject line",
|
|
"body_summary": "Summary of the email body content"
|
|
}
|
|
},
|
|
"invoice": {
|
|
"document_name": "Request for payment",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"invoice_number": "Unique invoice identifier",
|
|
"vendor_name": "Name of the vendor or seller",
|
|
"invoice_date": "Date of issue",
|
|
"total_amount": "Total amount due",
|
|
"currency": "Currency code (e.g., USD, EUR)",
|
|
"due_date": "Payment due date"
|
|
}
|
|
},
|
|
"presentation": {
|
|
"document_name": "Slide deck",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"title": "Title of the presentation",
|
|
"presenter": "Name of the presenter or author",
|
|
"slide_count": "Total number of slides",
|
|
"event_name": "Name of the event or meeting context",
|
|
"key_takeaways": "List of key points or summary"
|
|
}
|
|
},
|
|
"scientific_publication": {
|
|
"document_name": "Scholarly article",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"article_title": "Title of the article",
|
|
"journal_name": "Name of the academic journal",
|
|
"volume_issue": "Volume and issue number",
|
|
"authors": "List of authors",
|
|
"doi": "Digital Object Identifier (DOI)",
|
|
"abstract": "Full text of the abstract"
|
|
}
|
|
},
|
|
"questionnaire": {
|
|
"document_name": "Survey instrument",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"survey_title": "Title of the questionnaire",
|
|
"target_audience": "Intended audience description",
|
|
"number_of_questions": "Count of questions",
|
|
"response_type": "Type of responses (e.g., Multiple Choice, Open text)",
|
|
"instructions": "Instructions for the respondent"
|
|
}
|
|
},
|
|
"advertisement": {
|
|
"document_name": "Promotional content",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"brand_name": "Name of the brand",
|
|
"product_name": "Name of the product or service",
|
|
"headline": "Main promotional headline",
|
|
"offer_details": "Details of any discount or offer",
|
|
"call_to_action": "The primary action requested (e.g., Buy Now, Visit Website)",
|
|
"contact_info": "Phone number or website listed"
|
|
}
|
|
},
|
|
"form_10_k": {
|
|
"document_name": "Annual report of a company",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"year": "Year of the report",
|
|
"company_name": "Name of the company",
|
|
"ceo": "Name of the CEO of the company",
|
|
"net_income_millions": "Net income in million dollars"
|
|
}
|
|
},
|
|
"form_10_q": {
|
|
"document_name": "Quarter report of a company",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"year": "Year of the report",
|
|
"quarter": "Quarter of the report (Q1, Q2, Q3, Q4)",
|
|
"company_name": "Name of the company",
|
|
"ceo": "Name of the CEO of the company",
|
|
"net_income_millions": "Net income in million dollars",
|
|
"net_income_previous_year_millions": "Net income of previous year in million dollars"
|
|
}
|
|
},
|
|
"other": {
|
|
"document_name": "Unclassified Document",
|
|
"document_mime_type": "application/pdf",
|
|
"model": "gemini-2.5-flash",
|
|
"fields": {
|
|
"summary": "Brief summary of the document content",
|
|
"suggested_category": "A suggested category if none of the above fit",
|
|
"language": "Dominant language of the document",
|
|
"entity_names": "List of major entities (people, orgs) found"
|
|
}
|
|
}
|
|
},
|
|
|
|
"classification_config": {
|
|
"document_mime_type": "image/png",
|
|
"model": "gemini-2.5-flash",
|
|
"classes": {
|
|
"handwritten": "Document entirely handwritten. Process this before other classes",
|
|
"form_10_k": "Annual report of a company",
|
|
"form_10_q": "Quarter report of a company",
|
|
"resume": "CV or work history",
|
|
"file_folder": "Folder cover or label",
|
|
"budget": "Financial forecasts, spreadsheets, or internal expenditure authorizations. IF the document is a fill-in-the-blank form requesting or authorizing money (e.g., 'Contribution Request', 'Account Code', 'Amount: $'), classify as budget. EXCLUDE: Do NOT classify as invoice unless it is a final bill from an external vendor.",
|
|
"invoice": "Definitive demand for payment for goods/services already delivered. Look for 'Remit To', 'Net 30', 'Terms', 'Tax ID', or 'Please Pay'. If it is an external bill, it is an invoice, even if it has form-like boxes. EXCLUDE: Do NOT classify as budget.",
|
|
"scientific_publication": "Scholarly article",
|
|
"scientific_report": "Research findings document",
|
|
"specification": "Detailed technical, engineering, or architectural requirements. Look for section numbers (e.g., 1.1, 1.2), blueprints, materials lists, or strict technical constraints. IF it is a technical checklist or engineering form, classify as specification, NOT form.",
|
|
"questionnaire": "Survey instrument",
|
|
"email": "Digital message",
|
|
"memo": "Internal office correspondence. Look for a strict, block-style header at the top containing 'TO:', 'FROM:', 'DATE:', and 'SUBJECT:' (or 'RE:'). EXCLUDE: If the document starts with a conversational salutation like 'Dear [Name]', it is NOT a memo. Classify as letter instead.",
|
|
"presentation": "Official corporate communications or media press releases",
|
|
"news_article": "Clipping from a newspaper, magazine, or journal",
|
|
"advertisement": "Promotional content",
|
|
"letter": "Typed correspondence, usually external. Visually characterized by formal salutations ('Dear...', 'To Whom It May Concern') and sign-offs at the bottom ('Sincerely', 'Cordially', 'Best Regards'). EXCLUDE: If the document uses a rigid 'TO / FROM / SUBJECT' block header, classify as memo instead.",
|
|
"form": "Administrative data collection documents featuring blank lines, checkboxes, or fax cover sheets. EXTREME EXCLUSION: If the document contains prices, 'Account Code', authorizes spending, or contains technical engineering data, do NOT classify as form. Classify as budget, invoice, or specification instead.",
|
|
"other": "None of the above"
|
|
}
|
|
}
|
|
}
|