Semantic Annotations Protocol

This section is edited by Dott.ssa Susanne Müller with the collaboration of Dott. Alexander Auf der Heyde, Dott.ssa Annamaria Ducci and Dott. Giuseppe Marcellino.

  1. Dictionaries and ontology (Korbo2)
  2. Persons

2.1 Title (label)
2.2 Description
2.3 Unidentified
2.4 Signature

  1. Institutions
  2. Places
  3. Dates
  4. Artworks

6.1 Bildindex
6.2 Already Inserted vs To Be Inserted
6.3 Cycles
6.4 Engravings
6.5 Duplicates
6.6 Photographs
6.7 Unidentified
6.7.1 Lost artwork
6.7.2 Location unknown
6.7.3 Artwork photograph
6.7.4 Open note
6.7.5 Free resource
6.8 Artworks list for the Bildindex
6.9 Buildings as Artworks and Institutions
7. Bibliographic indications
7.1 Books
7.2 Reviews and Articles
7.2.1 Reviews
7.2.2 Articles
7.3 Manuscripts and drawings collections

  1. Dictionaries and ontology (Korbo2)

Korbo2 is a server bringing together external resources, mostly coming from LOD-compatible portals (Freebase, DBPedia, soon also Geonames, Bildindex, Europeana and VIAF), and resources created by annotators. Such resources are labelled according to the names of relevant authority files of the Deutsche National Bibliothek (DNB) or, alternatively, of the portal VIAF. Besides the title, resources are  standardised also on the basis of established ontological criteria.

      • Only the letters Ready For Annotations (RFA) can undergo annotation (interface: ‘Annotations’), t.i. letters whose text is in its final version
      • Text fragments are selected and annotated with the tool thepund.it.
      • To annotate editor will employ

– the simple triple tool
Lemma: to be used as Subject
Predicate
Object
– ‘templates’
‘Templates’ are used for more complex semantic annotations (eg. with the predicate ‘has address’ for places)
N.B.: To avoid creating duplicates, a check in KORBO is always recommended to verify if the resource already exists.

  1. If the resource is already present, then that is the one to be used
  2. if on KORBO no fitting resource is available, but an appropriate one is on Freebase, then

– ‘Copy and use’: The metadata is copied from Freebase without any modification

– ‘Copy in Editor’: The editor can edit metadata

  • Compilation place, compilation date, signature, recipient’s address, are NOT annotated: these data flow into KORBO through metadata.
  • If duplicates are still present during annotation, the choice falls on the older resource (the one with lowest ID), while the second occurrence flows into the “merging resources” list.
  • While working on a letter, annotators should turn the ‘status’ into ‘in-progress’. Once completed, the letter turns in ‘test’. After a final check the letter will be ‘validated’, such a change in the status involves a green tick by RFA.
  • Notebooks collect annotations and are shared within the working team.
  1. Persons

    Biblical and mythological figures are not annotated.

2.1 Title (label)

Names are normalised according to the DNB (copy and paste): last name, first name. Attention to medieval and Renaissance names (eg. Leonardo, da Vinci; Wolfram, von Eschenbach).
Where no normalised data is available in the Deutsche Nationalbibliothek, then these data should be sought in another authority file, ie. VIAF.
Normalised name for ladies usually includes both married and maiden name: eg. Brenner-Kron, Emma
Commas present in normalised names will be converted into dashes in the backend slug.

2.2 Description

Biographic resources are entities created in Korbo2.
To avoid generating confusion among persons with the same name, coincidence quite frequent in Basel, it has been decided to indicate in the field “description” also the wife’s surname separated with a dash from the husband’s (eg. ‘Burckhardt, Carl’ in the field “description” this entry becomes ‘Carl Burckhardt-Burckhardt’ or ‘Carl Leonhard Burckhardt-Ryhiner’ or ‘Karl Burckhardt-Iselin’).
The field “description” also provides the link to the corresponding entry of the DNB (or VIAF).

2.3 Unidentified

In the event non-identifiable people, places, etc., a new annotation will be added on Pundit using the appropriate predicate (person, place, etc.), thus creating a new resource in Korbo.
In particular:

  • Unidentified person: in the resource of Korbo, the text fragment is used in the label, while the field “description” will specify “not identified” or “not verifiable anymore”, together with the indication of the letter ID. eg.  LABEL: “due forlivesi veri rei”. DESCRIPTION: “Not identified. ID 12”.
  • Partially identifiable person (the only known element is the name, his/her family relationship with other already identified characters, etc.): in the field “description” will be added all available information together with the ID number of the letter.

2.4 Signature

Signatures are NOT annotated (the sender results from metadata and XML tags).

  1. Institutions

Under the category “Institutions” are considered authorities and organizations – publishers, universities or schools, public libraries, government agencies – also often appearing as senders.
In the triple
predicate: ‘Identifies Institution’.
title/label: follow the rules of the DNB for “Organisationen”. Eg. <Universität Göttingen>; not <Eidgenössische Technische Hochschule (Zürich)>
The valid name for an institution is its current and not the historical one. Eg. not .
Tabs are completed also with sources (Freebase, Wikipedia) and link to the DNB.
Institutions and architectural works: since the predicate has to be indexed, annotators must provide two entries, one for the work of art, and the other for the institution.
N.B. for the Tab Institution: No tick on the ‘artwork’ type in order to avoid duplicates.

  1. Places

  • To annotate Places, the appropriate template is ‘has address’. This template has 4 triples:

Identifies Place
Has Address
Has Latitude
Has Longitude
It is not mandatory to fill all the strings.

  • All places within the letter must be annotated (using the Normdaten of DNB). In case a same place is mentioned more than once, only the first occurrence will be annotated.
  • Geolocation (longitude and latitude) is provided only for indicated addresses. For geolocation refer to the site http://itouchmap.com/latlong.html
  • In the case of “Eingemeindungen” –  no more autonomous places, incorporated in larger cities or provinces, the choice falls on the modern name. Eg. Berlin, not Charlottenburg
  • Compilation places are NOT annotated (they result from metadata and XML tags). Exception: Institutions and places not resulting from Metadata.
  • In the field “description” is also added the permalink of DNB.
  • In the event of not identifiable places, a new entry will be added on Pundit using the predicate ‘place’ (new entry = new resource on Korbo). In particular: in Korbo the label contains the text fragment, while the field “description” specifies “not identified” or “not verifiable anymore”, together with an indication of the letter ID. Eg. LABEL: “Villa Bertha”. DESCRIPTION: “Villa Berta. Not identified. ID 161 “.
  1. Dates

For ages: beginning/final date and year, eg. for ‘Quattrocento’ will be added  1401-01-01-1500-12-31

  1. Artworks

6.1 Bildindex

Artworks are linked with a proxy to the corresponding Bildindex resource (BI: Bildarchiv Foto Marburg).

6.2. Already Inserted vs To Be Inserted

A distinction is made between the artworks already present with a tab on the BI (AI: already inserted) and those not yet included in the BI, that must be therefore catalogued (TBI: to be inserted). Burckhardtsource provides the BI with information about not yet catalogued artworks.

  • Annotations for AI

Title: Standardised following the DNB rules

Original URL: the URL of the BI for the object

Description: the ID of the object on the BI

Types: ‘artwork’, ‘bur-artwork-ai-bi’

  • Annotations for TBI

Title: Standardised following the DNB rules
Description: The categories to be added, followed by the character ‘:’ are:
ARTIST: (string describing the author)
LOCATION: (string describing the location)
DATES: (string describing date or date range)
ZEROS: (Tab number)
URL: (URL strings separated through commas)
DESCRIPTION: (single line string)
In each field, text has no new paragraphs.

6.3 Cycles

In cases of Cycles (AI) the BI assigns an object number (ID) only to the “mother tab”, so editors must refer to this tab for any detail.

6.4 Engravings

For engravings present in the BI, it must be indicated the sample copy preserved in Marburg. On the other hand, if specified in the letter, the sample chosen must be the exact one (engraver or publisher).

6.5 Duplicates

For Duplicates editors should choose the Foto Marburg document. Foto Marburg documents are those with no additional or with “Foto Marburg” source information in brackets after the identifier.

6.6 Photographs

The predicate for Burckhardt’s photos is: identifies foto
http://www.ub.unibas.ch/ub-hauptbibliothek/wir-ueber-uns/weiteres/jacob-burckhardt-edition/abbildungssammlung/

6.7 Unidentified

In the event of non identifiable artworks, annotators will create a new record of Pundit using the appropriate predicate and then create a new resource on Korbo (see procedure at section 2.3: Unidentified).

6.7.1 Lost artwork
For no longer existing artworks, the predicate remains the same (Identifies artwork), while a new resource on Korbo will specify in the field “description”: destroyed/lost (eg. Kirche Ingelheim, Dom von Novara)

6.7.2 Location unknown
For artworks with unknown current location, the predicate remains the same (Identifies artwork), while a new resource on Korbo will specify the location: unknown (eg. Tiroler Haus, Wachsmodell)

6.7.3 Artwork photograph
For artworks reproduced in photographs (Rubens Museum Berlin; ID 284) the entry will be double
Predicate: identifies artwork
Predicate: is related to +  picture URL

6.7.4 Open note
In case of additional information, annotators will create a free comment as ‘memo’. Its possible publication will be decided in the future.

6.7.5 Free resource
During researches on an artwork, annotators can create a ‘free’ resource in order to indicate the investigation stage.

6.8 Artworks list for the Bildindex

Artworks extraction, i.e. the creation of the list for Marburg, is carried out through a button on KORBO.

6.9 Buildings as Artworks and Institutions

Since the part to be indexed is the predicate, then buildings being both artworks and institutions will receive two distinct tabs (one for the artwork, one for the institution).
In the artwork tab: no tick on the ‘institution’ type, in order to avoid duplicates.

  1. Bibliographic indications

7.1 Books

Title: Citation APA of the tab Swissbib Basel-Bern (only author, date and title) (<http://baselbern.swissbib.ch/Search/Home>): Burckhardt, J. (1855). Der Cicerone. Description: Citazione MLA della scheda UB-BS
Burckhardt, Jacob. Der Cicerone. Basel : Schweighauser’sche Verlagsbuchhandlung, 1855.
URL of the DNB, or, if present, a perma-link for the book in the Staatsbibliothek Berlin, in the Staatsbibliothek München, or in the Bibliothèque Nationale (Gallica).
Indications of pages: template “cites page”
No images has to be added.
In the event of a generic citation of a review in a letter, annotators will create an entry on Korbo following the same criteria used for books, but those used in the label are the Normdaten: if over the years the review has undergone a name change, the name adopted will be the one under Titel des Werkes of the DNB.
In the event of a generic citation of a Greek or Latin work (eg. Pharsalie), annotators will create an entry on Korbo and indicate collection Loeb as reference edition. Additionally, will be included also the permalink to the Staatsbibliothek zu Berlin or Staatsbibliothek Munich.

7.2 Reviews and Articles

7.2.1 Reviews
Title: <Burckhardt, J. (1845). Rev: G. Kinkel, Geschichte der bildenden Künste>
Last name, abbreviated name (publication year). Rev: abbreviated name, last name, shortened title (no subtitle). Types: the only one used is “Journal”.

7.2.2 Articles
The citation criteria adopted by the Swissbib Basel-Bern for books are accepted here for both title and description. In addition, annotators will record in the field “description” also the review name in inverted commas, followed by the number, the year in parentheses, and the pertaining pages preceded by pp. Moreover, when possible, annotators will add also the link to DigiZeitschriften. Eg. Bode, Wilhelm von. Die Ausbeute aus den Magazinen der Königlichen Gemälde-Galerie zu Berlin. Mit vier Hochätzungen, “Jahrbuch der Königlich Preussischen Kunstsammlungen”, 7 (1886), pp. 226-243.
Types: the only one used is “Journal”.

7.3 Manuscripts and drawings collections

Both label and description indicate:
Place
Institution
Cataloging
In case of missing cataloguing, the chosen name will be the one under which the code is universally known. Eg. “Paris, Musée du Louvre, Département des Arts Graphiques, Codex Vallardi”. The field “description” indicates also a permalink for the code or, in case this is not available, then a link to the manuscript (eg. the Codex Vallardi: link to Wikipedia).

 

Last updated 2015-09-03