#TEXT ENCODING INITIATIVE TEI PDF#

It is actually not always straightforward to achieve 100% well-formedness because the PDF input documents can be very difficult to parse, the content very noisy, not always fully recoverable, and the Grobid parsing can include many errors. Grobid ensures that all the generated results are well-formed XML. We recommand to use the RelaxNG schemas, more robust, easier to use, etc., and avoid W3C schemas which are recurent sources of problems. W3C XML schemas, under grobid-home/schemas/xsdДЄ documentation of Grobid customized TEI is available grobid-home/schemas/doc/Grobid_doc.html. XML RelaxNG: grobid-home/schemas/rng/Grobid.rngД¬ompact notation RelaxNG: grobid-home/schemas/rng/Grobid.rnc The Grobid ODD specifications can be found under grobid-home/schemas/odd/Grobid.odd.Дifferent schemas generated from the ODD specifications by ROMA are available in the repository: Having ODD descriptions, it is possible with a tool called Roma to generate automatically customised TEI schemas (xsd, relaxNG, etc.) and some documentation. To quote the TEI Guidelines: 'it is almost impossible to use the TEI scheme without customizing or personalizing it in some way.'ДЇor customisation, the TEI community has created a specification language called ODD ('One Document Does it all') to modify the general TEI schema. A customization adapts and constraints the richness of the TEI to a well scoped and tuned schema. The TEI define a very large and rich set of encoding possibilities, often redundant taken as a whole. The schemasДЄny usages of the TEI for a particular sort of documents suppose a customization of the TEI to their specificities.

The TEI appears therefore as a suitable document model and XML implementation for the Grobid structured document extraction. The Guidelines define some 500 different textual components and concepts (word, sentence, character, glyph, person, etc.), defined and refined over the last 25 years. TEI allows very precise text annotations while maintaining human readability. Among the different XML standards, the TEI (Text Encoding Initiative) is probably the most comprehensive and mature encoding standard for machine-readable texts. Derived from SGML, XML has been designed exactly for that. A document model has to cover multiple embedded/recursive sections, reference mechanisms, lists, etc.

Representing a complete document requires a much more sophisticated data model.

#TEXT ENCODING INITIATIVE TEI FULL#

In other terms, Grobid aims at extracting and restructuring the complete full text of a document.

However, the goal of Grobid is complete PDF document ingestion. Using tabular or POJO representations is adapted to bibliographical records, which are relatively flat and simple structure.

Text encoding initiative tei