Applicant Authenticity Analyzer AI-powered Apify actor that reviews job application documents (PDF, DOCX, or plain text) and determines how likely they are to be authentic. The actor extracts readable text, runs a fraud and AI-generation assessment through an AI analysis service, and returns a verdict, justification, and trust score from 1 (clearly fake) to 10 (highly authentic). ## 🚀 Key capabilities - Accepts pasted text or publicly accessible file URLs (PDF, DOCX, TXT) - Converts documents to text automatically using embedded parsers - Evaluates authenticity, highlights risk signals, and suggests verification steps - Produces JSON output with verdict (likely_real or likely_fake), score, justification, and confidence - Uses a tuned AI prompt with temperature controls and optional custom system instructions ## 🧾 Input | Field | Type | Required | Description | | --- | --- | --- | --- | | maxItems | integer | | How many applications to process. Free users are limited to 100. Paid users can process up to 1,000,000. | | applicationText | string \| string[] | | Raw application text pasted in input. | | applicationFileUrl | string \| string[] | | URLs pointing to PDF, DOCX, or TXT documents. | | jobRequirements | string \| string[] | | Optional list of requirements to evaluate job fit. | | jobResponsibilities | string \| string[] | | Optional list of responsibilities to evaluate job fit. | Provide at least one applicationText or applicationFileUrl. If you supply job requirements or responsibilities, the output will add a jobMatch percentage showing how well the application aligns with the role. ## 📦 Example input json { "applicationFileUrl": [ "https://example.com/resume-john-doe.pdf", "https://example.com/cover-letter-jane-smith.docx" ], "jobRequirements": [ "8+ years of backend development experience", "Expertise with PostgreSQL and distributed systems" ], "jobResponsibilities": "Own the design and delivery of backend services for the hiring platform." } ## 📤 Output Each dataset item contains: json { "sourceId": "file-1", "sourceType": "file", "reference": "resume-john-doe.pdf (application/pdf)", "verdict": "Likely fake", "score": 3, "confidence": "high", "jobMatch": "42%", "summary": "Application shows strong signs of templated AI generation with contradictory dates.", "justification": [ "Cover letter repeats identical phrasing across multiple paragraphs.", "Resume claims conflicting employment dates between sections." ], "riskSignals": "Boilerplate language with minimal personalization; Inconsistent job titles and timelines", "consistencyChecks": "Two roles overlap for 18 months without explanation", "nextSteps": "Request original references to validate employment dates" } When the actor cannot process a document, the dataset will include an item with error describing the failure. ## ⚠️ Notes - Documents must be reachable via HTTPS URLs; authenticated downloads are not supported out of the box. - Legacy .doc files are not supported; convert them to .docx first. - Always comply with privacy policies and handle applicant data securely.