- Extraction recovers text and document structure.
- Segmentation identifies sections and repeated entries.
- Entity extraction maps text to fields such as employer, title, dates, and skills.
- Normalization can connect variants to a shared concept.
Stage 1: document extraction
The system opens the uploaded file and recovers text plus whatever structural clues the format exposes. DOCX provides XML elements. PDF extraction may rely on positioned glyphs, font mappings, tags, or OCR output.
Stage 2: section and entry recognition
The parser looks for boundaries such as Work Experience and Education, then groups nearby lines into records. Repeated patterns help it decide which date, title, and employer belong together.
Stage 3: entity extraction
A commercial parser can return a wide field set. RChilli documents fields for names, contact data, employers, job profiles, dates, education, certifications, and skills. The exact schema belongs to the parser and its customer configuration.
Stage 4: normalization and profile creation
Raw terms may be mapped to normalized titles or skills. The recruiting system can then place fields on a candidate profile, make them searchable, or pass them into another matching feature. Recruiters may still see the original resume beside those fields.
Where errors enter the pipeline
| Stage | Example error | Resume-side check |
|---|---|---|
| Extraction | Characters are missing or reordered. | Search and copy all text from the export. |
| Segmentation | Projects are mistaken for jobs. | Use distinct, conventional section headings. |
| Entity mapping | A date is attached to the wrong employer. | Keep related fields close and use a repeated layout. |
| Normalization | A niche skill is mapped too broadly. | Keep the precise term and add a common expansion when accurate. |
Sources used on this page
- Resume Parser FieldsRChilli Documentation
Shows the structured fields a commercial parser can return, including experience, education, contact details, and skills.
- Understanding resume parsingLever Help Center
Describes the fields Lever adds to a candidate profile after parsing.
- Manually add a candidate or prospectGreenhouse Support
Advises recruiters to verify parsed fields and notes that embedded images or some layouts can produce incorrect imports.
- Structure of a WordprocessingML documentMicrosoft Learn
Documents the paragraph, run, and text structure inside DOCX files.
- What you may be missing when you search PDF documentsPDF Association
Demonstrates how PDF content order can differ from the order visible on the page.