How to extract data from old birth certificates
Historical birth certificates — especially those pre-dating the 20th century — contain far more genealogical information than modern ones. Beyond the name and date, many include full parentage, parents' place of origin, grandparents, witnesses and even the family's hometown. Knowing which fields to look for and how AI identifies them can greatly speed up research.
What a historical birth certificate contains
Civil birth certificates (in Brazil from 1888, in many European countries earlier) and parish baptism records (17th-19th centuries) have different structures, but generally share the following fields with genealogical relevance:
| Extracted field | Example | Note |
|---|---|---|
| Registered name | Person A | Full name, sometimes with nickname |
| Date of birth | February 3, 1891 | May differ from registration date |
| Registration date | February 10, 1891 | Date the entry was recorded |
| Place of birth | Sao Leopoldo, Rio Grande do Sul | Municipality, village or farm |
| Father | Parent A | Name and sometimes occupation and place of origin |
| Mother | Parent B | Maiden name may appear |
| Paternal grandparents | Grandparent A and Grandparent B | Common in 19th-century certificates |
| Maternal grandparents | Grandparent C and Grandparent D | Especially in European documents |
| Witnesses | Witness A and Witness B | Often neighbors or godparents |
| Recording official | Registry official | Responsible registry or parish |
| Entry number | Entry no. 124, Book A, fol. 38 | Archive reference |
Difference between civil certificate and baptism record
Civil birth registration (in Brazil from 1888 under Law No. 1,144; earlier from the 1870s in some regions): recorded at a government registry office, required by law. The structure varies by country, state and period. Older certificates tend to have more handwritten fields, with significant variation from one registry to another.
Parish baptism record (before civil registration, and concurrent with it to this day): recorded by the priest in the parish baptism book. In European immigration regions, often written in German, Italian or Latin. Contains godparents, which are absent from civil certificates. It is the primary genealogical source for births before civil registration was established.
For genealogical purposes — especially in Italian or German citizenship by descent research — it is common to need both types: the parish baptism for the oldest ancestors and civil certificates for more recent generations. Each source fills gaps the other leaves open.
How AI identifies these fields
Extraction works in two steps. First, the document text is read via OCR or a computer vision model for handwritten manuscripts. Then, linguistic patterns typical of certificates indicate the position of each field in the text.
Documents with a more standardized structure — printed forms with handwritten blanks — have higher extraction rates than documents written entirely in free-flowing text. Many 19th-century certificates follow a narrative format where a clerk wrote the entire entry as continuous prose, making field boundaries harder to detect.
KinProof compares found surnames against a database of known variants, allowing identification of the same surname in different spellings (e.g., Surname A/Variant A, Surname B/Variant B, ExampleSurname/ExampleSurnameVariant1). This surname variant matching is particularly important for German-Brazilian records, where spelling was often adapted to local conventions.
Important limitations
- ⚠Certificates with very cursive or deteriorated handwriting may have incomplete fields in extraction.
- ⚠Grandparents' and great-grandparents' names are frequently abbreviated or omitted — extraction reflects what is in the document.
- ⚠AI does not infer absent kinship relationships: if the father is not listed, the field is left blank.
- ⚠For Italian, German or Portuguese citizenship processes, AI transcription does not replace reading by a registry, consulate or qualified genealogist.
- ⚠Documents with erasures or corrections made by the clerk may produce two values for the same field.
- ⚠Baptism certificates in Latin require familiarity with ecclesiastical formulas for correct interpretation.
Upload your historical birth certificate as a PDF and see which fields KinProof can identify automatically.
Test with my certificate