The project consists of several complex stages with different objectives, from collecting numerous data, organizing information appropriately, investigating them based on available knowledge and finally transforming the analysis results into visual representation. Therefore, we are using the sequential unified process throughout the whole project, specifically along with modern agile development approach to gain velocity, progressive deliverables, and iterative evaluation for work construction phases (data analysis and website implementation). The significant advantage of apply agile practice in those phases is “tangible deliverables” at the end of each iteration (sprint), which allows assessing work quality and motivates improvement during the development process.
Figure 1. Development process methodology
Paintings, which are the core element of this project, will be collected from different sources available online. The most efficient source could be investigating from famous painters (who are mostly members of scuola) during the Renaissance Age, such as the revolutionized painter Giovanni Bellini, or Giorgione, who is rightly known as the first really big “name”. Their artwork inspired notable followers such as Titian, Tintoretto, Paolo Veronese and Lorenzo Lotto, etc.
Important information also comes from historical documents about Venice organization back in the 16th century, as a reference source for the power represented in paintings being analyzed. Even though a majority of historical references about power in Venice can be found in the book Civic Ritual in Renaissance Venice (Muir, 1986), there is room for more researches over various online sources available.
These are some more data sources that we will mainly target to (but not limited in) during the collection process:
- Published papers from academic seminars about Venice in Renaissance
- Universities’ publications
- Reputable institutions of Humanities Science
- Websites about Renaissance art
Organizing data collected in the previous selection process requires the most effort to accomplish. Data aggregation, after needed information has been gathered, is a complicating phase that includes selecting appropriate data to be stored, verifying reliable data sources and then merging to the existing archive model.
Figure 2. Data organization steps
Since our objective is visualizing the power through events being described, the created time and real event in the paintings are the most fundamental information that needs to be clear and precise. While the time is crucial to trace back to the situation in which the painting’s author created his artwork, exact event name (and time) would allow further investigation about social conditions and the organization of Venice power at that time.
Moreover, since images are unstructured data, they are not efficiently stored by traditional relational database model. Instead, each image is represented by an object that contains all necessary information for a standard painting, such as: name, created date, author(s), event, location, URL, etc. In the future, each painting might be divided into multiple segments that depend on different interest areas of analysis and will be discussed in details in later phases.
Below is the process of extracting and populating data into database:
- Partition each category of paintings
- Import images to cloud database, obtain individual URL
- Centralize information about each painting
- Convert all data into consistent form
- Verify the whole database and fix issues (if any)
Analysis of data is a huge practice and has hundreds of approaches to follow. Information about paintings is categorized into 2 levels:
- The first abstract level contains three big sections: Physical factor, Visual factor and Property of people described in the painting.
- The second level includes subclasses of each group above. For example, physical factor can be observed by position a person in the painting, relative position compared to others in the crowd, what he is doing in the picture, the number of objects with the same characteristics as him, and so on.
Figure 3. Hierarchy of power relevance in paintings
Data analysis also involves the selection of visualization model to be applied to a particular painting appropriately. There are several types of visualization  that can be used as the fundamental concept for power represented in paintings:
- Tree/Hierarchical: particularly useful for the purpose of representing power, in which levels of organization could be presented using dendrogram, radial tree, hyperbolic tree, tree map, partition chart, etc.
- Temporal: data is organized in time series, for example, the changes of Venice power over time between two paintings. Temporal visualization includes timeline, Gantt chart, stream graph, arc diagram, and so on.
- Multidimensional: involving various types of chart to represent statistics of objects in the painting such as: the ratio of politicians/civilians, years on duty of the doges, Venice demographic factors, …
- Network: different objects in the painting can be connected and related to each other using matrix map, node-link diagram, dependency graph/circular hierarchy, subway/tube map, etc.
Historical paintings, due to their complex of coloring condition and large size, usually occupy a large volume of storage. Therefore, we aim to utilize the powerful capability of cloud computing to store images of artwork so that they are completely separate from the analysis system and web application. It also enhances the scalability of the system for future expansion.
Figure 4. Architectural design of the system
Detailed project plan is described in a separate post that can be accessed here.