Data Model
A DB2 database, enote, was created and stores all scanned music scores including its meta data and its feature vectors, which represent
the writing characteristics.
In order to logically separate the data, all mentioned types of data are stored in the following database schemas:
- DICT: The database schema DICT is used to define tables which store
characteristics of handwritings. They represent the feature base.
- FEATURES: The features as well as the results from the semiautomatic
analysis are stored in tables from the database schema FEATURES.
- METADATA: This database schema serves to define tables for storing source
descriptions and library data of the digitized handwritten music scores.
- IPFV: The features of handwritten music scores (especially note heads and note
lines) are stored in tables of the database schema IPFV.
In the following, all tables will be explained in detail. For further information, please have a look at the documentation,
which only exists in German.
The characteristics of handwritten music scores are represented by a tree hierarchy, called feature base. The following tables represent this tree hierarchy:
- Node_Type contains all existing node types
- Nodes stores data about handwriting features
- Distances contains the distance between two nodes and thus, the distance between two handwriting characteristics
For detailed attribute definitions of each table, look at relational data model of the schemas DICT
The schema FEATURES contains tables which store results of handwriting analysis. The result of each handwriting analysis is a feature vector that
stands for a set of predetermined handwriting features. A feature vector can belong to one or several music manuscripts as well as to one writer or different
writers. In the database schema FEATURES, the following tables are defined:
- Featurevectors represents the feature vectors
- MMS_FV stores the relations between feature vectors and music manuscripts
- FVvalues contains all values for a feature and associates them with the corresponding feature vector.
The relational data model of the schema FEATURES delivers detailed informations about the attribute
definitions.
This schema contains the following tables which store data about the written music scores like images, general information of the note sheets, the composers
and the scribes:
- In Music_Manuscript, all general library information about the music manuscripts are stored. The official library seal indicates
a library and together with the library signature of handwritten music scores, it is used as primary key of the table.
- The table Music_Work contains all music compositions that belong to a music manuscript including its title and comments.
All other information is stored in further tables and referenced by foreign key.
- In the table Composer stores the relations between composers and music compositions.
- The table Text_Author stores the relations between music compositions and text authors who have written the text to the
corresponding music composition.
- Incipit stores the first notes at the beginning of a music composition.
- Manuscript_Section stores information about handwritten music scores' sections like text comments, time and place of the creation.
- In Music_Manuscript_Page, features about a page are stored including paper characteristics like format, height and width.
- Page_Images contains the digitalized image of the handwritten music scores as well as information about it.
- Music_Manuscript_Scribe stores relations between the music manuscript pages and their scribes.
- Libraries contains libraries which are identifiable with its library seal
- Music_Score_Collections stores information about the music manuscript collections. The catalog "Beschreibender Katalog" contains
handwritings, compositions and composition sets of several composers. The collection "Anonymus" contains handwritings and compositions of unknown
componists.
- Scribes stores scribes including information about his work period, his residences and contacts.
- Music_Works stores general information about a music composition like its title, required instruments and its key.
- In Text_Authors, text authors including their names and their residences are stored.
- Composers contains the composers including their names and their residences.
- In Incipit_Types, the types of incipits are stored. Up until now, the only existing types are "Content" and "Paper Description".
Incipits are the first notes at the beginning of a music composition.
- The table Tones stores the keys of the music compositions. Keys A, B, C, D, E, F and G are available.
- Roles contains music instruments which are required in order to play the music composition.
- Section_Types stores all possible types of music scores' sections. The following types currently exist: score,
score fragment, voice and envelope.
Take a look at the relational data model of the schema METADATA.
In this database schema the following tables are defined. They store the components of a staff like note lines, bars, note heads and note stems.
- Page_Image_ROI stores the sections of a note sheet to analyze.
- Staff_Lines contains the note lines on the note sheet.
- Note_Head stores the note heads of the note sheet.
- Note_Stem stores the note stems on the note sheet.
- Bar_Lines contains the bar lines of the note sheet.
The relational data model of the schemas IPFV gives additional information.