bald_splash

Lomen is a research project aiming at exploiting the richness of digital archives, stitching up the relationships between entities and providing a visual access to data. It is a web-based platform that allows users to explore contents in a non-linear way, identifying patterns and fostering insight. The platform also aims at the sedimentation of several levels of information, through direct linking to archive entities, such as projects, artifacts or individuals involved. Curators are also given the possibility to elaborate theme-based paths, providing to users a different and original access to the underlying data.

Background

The digitization of archival collection created new ways to open contents to a wide public, creating new challenges and opening the field for new experimentations.

Providing access to digital archival collections is becoming a primary concern for curators; cultural institutions are becoming aware that simply creating an online repository for the archive is not enough to reach a vast public, and that availability doesn’t necessarily bring access. As a result, there is a growing need for the design of engaging interfaces, capable of supporting the exploration of digital archives.

Lomen addresses two main issues. The first one regards the creation of a visual interface for archives, able to represent different kind of entities; identifying a possible interaction that allows users to visually shift between the diverse types of entities.

The second one concerns the provision of a context, by allowing archive’s curators to sediment several layer of information upon the database and building thematic paths linked to archive’s contents.

The Baldessari archive

The Lomen project has been applied to the historical archive of Luciano Baldessari, a notable Italian architect. His archive belongs to three main funds, owned by different institutions. The main fund is managed by the LADA laboratory of the Politecnico di Milano, and contains Baldessari’s technical drawings and his private mail. The other two archives, owned respectively by the CASVA center of the Milan municipality and by the MART museum contain photographs and paintings. The archive is divided by architectural projects: each project is described by the geographic location, the actors involved, the chronological extremes of the realization, the related bibliography, the drawings and the correspondence regarding the project.

The aim of the project was to identify how to provide a public, web-based access to the archive able to explicit the relationship between entities, creating a visual interface allowing users to explore the archive. Also, the application should have been flexible enough to handle inconsistencies in metadata and possible modifications of the database. A Further objective of the project was to provide to curators a backend to modify existing records and create new ones.

Lomen is structured around entities extracted from data and identified as the main elements of the corpus. For every entity we defined a set of suitable views, each one focusing on one or more characterizing attributes. Moreover, for every entity we defined a group of custom filters allowing the user to refine the corpus and to narrow the set of entities being shown in the current view.

The main advantage of this approach is that it enables an efficient and agile navigation throughout the corpus, allowing the user to see the same entity under different points of view.

Views

In the Baldessari archive we identified 3 main entities: projects, documents and people. Projects are pivotal elements, as the other two types always refer to them. These are the related views:

Projects

  • list: this model offers a simple, general and efficient way of exploring the projects in the archive, providing a small thumbnail of a related document and a brief textual description.
  • gallery: the gallery view follows a visual driven approach to the projects exploration, made by a grid of stacked items where every cell is occupied by the thumbnail of a project and its title. Arranged by chronological order, the grid view allows the user to see at a glance the different styles used by the architect in his projects, find similarities and contrasts or look for a work characterized by a particular visual style.
  • timeline: the timeline view focuses on the chronological ordering of the projects: the story of Baldessari’s documented works is represented as an horizontal scrollable panel in which every item is denoted by a marker and its reference dates. This visualization reveals to be crucial to analyze the development of the architect’s career over time and find periods characterized by similar activities or projects types.
  • map: this view offers a geographic representation of the projects distribution. The user may interact with the map by zooming in and out and moving the region of interest. Depending on the zoom level, projects located in close areas are represented as clusters which can be exploded and seen as single items. This visual tool is useful to see places and countries where Baldessari worked the most, or to find out projects in uncommon localities with respect to the architect’s life.

In any of the above mentioned views the user can select a single project and see its detail page, which describes in depth its metadata and offers a gallery of all the related documents.

Documents

  • list: every item of the list is composed by a thumbnail of the document, followed by the related metadata: related project, signature, technique, description, scale, instrument, support and notes. Some of the metadata strings act as quick filters.
  • gallery: a grid of stacked documents, each one showing a thumbnail, the signature of the item and its related project

People

In the case of the actor entity, the user is principally interested in seeing what were the relationships between Baldessari and other people and organizations related to him and his works. Based on this assumptions, we implemented an interactive network graph where the nodes are divided in two categories: actors and projects. An actor node can only be connected to a project node, and vice versa. Each node size depends on the number of edges connected to it. The resulting visualization, developed with a force directed drawing algorithm [3], allows the user to explore the collaborations of the architect highlighting the most important relations and projects of his career.

The user can interact with the graph by zooming in and out, moving the center of the view and selecting a particular node. In the last case, only a subset of the relations is highlighted, starting from the clicked element and going through the connections up to the third level.

This visual tool enables a deep exploration of the relations: recognizing the most important projects and actors is easy, and so is the discovery of small, isolated ones.

Filters

For every entity defined above, we built a set of related filters to narrow the research and dynamically create a subset of data. These filters can be stacked one above the other resulting in a chain where every filtering is applied after the other with a logical AND fashion (e.g. a project view filtered by name and date will return entries containing the search string in the title and starting from the specified starting date). Moreover, filters can be dynamically added or removed, without the need of reloading the page.

We defined three global filter typologies, each one suitable for a specific kind of data:

  • Search box: this type of filter is intended for free text metadata such as titles, descriptions, etc. The filtering is implemented in an instant search fashion and aided by autocomplete choices, which help the user in finding the exact term he was looking for.
  • Multiple choice checkboxes: this filter has been designed for data belonging to a closed, narrow set. It is used for example to filter the documents by their related activity (architecture, painting, sculpture). Every checkbox is added to the search with a logical OR fashion (e.g. if we check architecture and painting the view will return all the documents related to architecture and/or painting).
  • Sliders: this filter type is used to select a range between two extremes which may be dates, geographical coordinates, etc.

The platform can be extended to use new kinds of filters, depending on the need of the corpus to which it is applied.

As we defined different views for each entity, we thought the filters as consistent elements throughout the visualizations. In this way the user can create his subset of the corpus related to a specific type of data and analyze it in different contexts. For instance, if we set a sliders filter on the projects map, we may see the same results in the timeline just by switching the view.

Thematic Paths

As stated before, one of the main goals of the Lomen platform is to allow the user to extract new information starting from the initial data, find interesting patterns and clusters of homogeneous entities and create new valuable content.

In order to help this process, the Lomen platform offers to archive curators a tool to create thematic paths based on data present in the archives and shown in the views. Each thematic path is composed by an arbitrary number of steps, and each step is composed by a title, an image and a describing text. The steps are shown in chronological order in a timeline based visualization similar to the related projects view described before, and the user may browse them freely scrolling the horizontal panel.

The entities previously defined can be referenced in the steps of the thematic paths by using a language similar to the wiki markup language used by wikipedia. This syntax is simple enough to be understood by users with no technical background, and may result familiar to students, researchers and expert users in general due to its similarity with the wikipedia one.

An additional advantage of using qualified links is the possibility to show a visual summary of the current step of the thematic path. At the side of the descriptive text a set of boxes shows the user all the entities cited in the page: projects, actors, documents and other themes. In the case of documents, instead of a textual reference a thumbnail for each item is shown. This synthesis may be used to have a context at a glance of the arguments discussed or to find quickly some interesting referenced items.

Technical Details

Lomen is a web platform developed using Django for Python as web framework. Its ease of configuration and orientation to rapid development allowed us to focus on the design of the interfaces, sparing time on the programming side. The structure of the system relies heavily on the digitalization of the archive, which was modelled following an entity-realtionship fashion and realized using SQLite. This database was chosen for its integration with Django and for the nature of the Baldessari use case, which was composed of a relatively small number of entries.

The front-end is by far the most complex part of the platform: we leveraged the latest technologies in order to build dynamic, agile and complex views. The tools we used include:

  • HTML5 and CSS3
  • Jquery
  • Angular.js (data binding)
  • D3.js (data visualization)
  • Leaflet (interactive maps)
  • Masonry (dynamic layout)

A big effort was spent in the integration of these frameworks into an organic and transparent interface, using cookies in order to maintain the filtering chosen by the user over the data across different views.

If you’re interested in using the Lomen platform, contact us at info@densitydesign.org