Reaction: Automating the Design of Graphical Representations.

The research paper explains the importance of differentiating the work of presentation from the application development. Always when dealing with large sets of data samples it is often tedious for development of applications to also take care of the finest details of effectively presenting the data. Presenting the data in the most fashionable way is the need of the hour with the availability of high powered graphical hardware systems. Building software models for such hardware systems with ever increasing features is a tricky task. An application might have the best possible design structure, but if the user interface of the system fails there are high chances of the system not being accepting in good number and visualization is one of the best available options in building a better UI.

The paper explains the approach of the authors in developing an application independent presentation tool that automatically designs effective graphical designs of graphical data. The assumptions taken into consideration in developing this model is that graphical representations are sentences of graphical languages with precise syntactic and semantic definitions. They have used artificial intelligence techniques to implement a prototype, A Presentation Tool (APT), based on composition algebra and graphic design criteria.

The paper focuses on automating the design of 2D static presentations of relational information and explains that expressiveness and effectiveness are two significant parts of building a better graphical model. The paper also explains in brief about the previous work, separating presentation data from application, in the associated field of automating graphical data by quoting references of AIPS system and the automatic animation scripting & rule based layout of node link diagrams. Content issues were the primary focus for the work on the two systems mentioned (VIEW and APEX). Other models that were referenced here were the BHARAT and the Beach model. The author explains the structural design of the framework using a set of automobiles and other associated features of the automobiles as an example using 2D models( bar charts and pie graphs). Bar charts are used when there is a monochrome information and pie graphs are used when multi-color set values can be included in the data set.

Things taken from this reading:

i. Effective way of expressing relational information in a graphical approach.

ii. Cases when bar charts and pie graphs are to be used.

iii. Real importance of separating presentation from application.

Things that look vague in the paper:

i. Most of the data sets in every day use are not always relational information. How should tools like these be used in such cases.

ii. What graphical models/languages are to be incorporated for specific data sets. (relation between graphical models that are to be used to form the graphical language)

Something that can be done to improve the work done on this is that, include feedback mechanisms to take in data at various other intervals over the course of the development.