With the support of the Media Futures programme Start up Meets Artist, in the past five month I collaborated with Studio Ramco van Bladel in the development of Biblio-Graph, a software for collective data aggregation for the GLAM sector [galleries, libraries, archives, museums].
Biblio-Graph works as a semantic layer in combination with the existing databases of archives and libraries.
Through Biblio-Graph readers, researchers and staff can interfere, manipulate, share and reuse data without interfering in the original database of libraries and archives. It will allow users that have domain specific knowledge, but not necessarily have technical knowledge, to make triples while reading, and to add linked data to the database through a semantic model.
This is achieved by combining two micro tools, text annotation and voice recording.
Text annotation and voice recording are not new technologies, on the contrary there are several available in the market, that allowed us to test and experiment to further develop a tool that is tailored to our needs.
The text annotation tool offers 6 different labels that may vary according to the library. In the case of de Appel, we use Place, Period, Entity [person, organization], Objects [works, books], Events, Terms and Phrases. The set of available concepts generate movement and circularity, data from the content of a book will relate to the metadata of other books. And every new piece of data will affect different regions of the graph, and generate inference. All declared relationships reveal implicit relationships, producing new knowledges within the collection.
When a word is annotated, all the other occurrences of the same word will also be automatically annotated. This will allow for the reader to easily see how a term, concept, person, organization or place appears in the rest of the collection, and how they relate. Through text mining tools, the reader can access all the paragraphs in which a term occurs, expand or concentrate the selection.
The voice recording is intended for recording fragments of texts. From a sentence to a paragraph at a time. Each recorded fragment is added to the metadata of the book and becomes immediately available for playing. Content of audio fragments are also annotated. Allowing for queries to be played as audio.
As a result of aggregating data, Biblio-Graph offers Play, a menu of functions that allow for mapping, visualization, navigation and browsing of the entire collection, as well as annotated and recorded data in different levels of granularity, through bookshelves, spines, covers, pages, recordings and annotations, and in relation to different periods, places, people and organizations, events, terms and phrases.
We believe that every library has a story to tell. For this reason, the development always starts through the databases of library and a study of the history of the institution. This will reflect in the schema. The semantic model is based on an upper/mid-level ontology and it is intended to support knowledge graphs for cultural libraries and GLAM. Based on this general framework we can further develop a model that is specific to the cultural library domain so that librarians, artists and researchers can gain a better insight into the content of library collections.
The Biblio-Graph ontological model aims to depict the minimal set of upper and mid layer concepts, relationships and definitions required to represent and express the vaster universe of concepts present in the realm of cultural libraries. It can be pragmatically developed based on the input of librarians, researchers, artists and scholars from different disciplines.
Bibliograph was selected as best startup and best artist in the “Startup meets Artist track”.
You can find more information here.