The evolving demands of the market and the increasing competition force companies
to support their business plans with tools that help managers to make decisions
in a rapid and more effective way. Presenting data in a convincing and understandable
way requires a lot of work in particular when data changes dynamically. The research
activity in this section of IVU Group is focused in developing Visual Data
Mining tools, an emerging area that is interested in performing explorative
data analysis and mining. Visual Data Mining refers to methods for supporting
exploration of large data sets by allowing users to directly interact with
visual representations of data and dynamically modify parameters to see how
they affect the visualized data, in few words,
it allows users to put their hand on data in order to improve the decision
making process.
The graphical presentation of the data allows users to discover
specific patterns, as well as new and useful properties in the data, their correlations,
and also detect possible deviations from the expected values.
In this section are presented some tools developed by the IVU Group. Each
tool is presented with a short description that describe what the tool is and
which are its main goals. Then there is a link to a more detailed page containing
references and screenshots of the interface.
For any comment
please send an e-mail clicking
here
-
A user do not want to make queries on a database
or wait too long in order to have results of a data mining process. Typically,
when dimensions for analysis are identified the user want to push a button
and see data. DaeClient ask the identity of the
user, retrieve information about the analyses and tools associated with the
specific user and shows them with a simple interface. The user can then select
the analysis he want to perform.
-
DaeVET can connect to a database and show the structure
of a data source in order to prepare data to
perform later an analysis. The intended user Improving
this task will result
in a reduced time to acquire data, hence, there is an improvement in the knowledge
discovery process.
DaeVET has been made for an expert user.
-
DaeTL, inspired to the Table Lenses technique, provides users with an overview of the data they are analyzing.
A fisheye effect allow them to zoom into elements of interest without loosing the overview.
Automatic zooming to show the details while moving the mouse is one of the feature of the tool.
It is possible to zoom intervals instead of zooming on single rows.
This tool automatically finds the right way to visualize the different data type
without asking the user how to show them. Hence the user doesn’t need to configure the tool since the data are
automatically visualized depending on their type.
-
DaeQP is a tool inspired to the Query Preview technique and provides users with
rapid overviews of information about objects of the user domain in order to
perform appropriate data analysis along different attributes (dimensions).
The power of this visualization is that
permits to perform a direct comparison between several different variables.
The overview shows the data distribution along the selected attributes.
Then, the use dynamic queries and query previews supports efficient query formulation.
-
ARVis visualizes Association Rules (AR) using graphs and allows the user to interact with them.
ARVis presents the overview of the dataset, then permit the user to zoom, filter and provides the details on
the user demand. The main user for ARVis is the data miner, and the goal is to support the user in explorative data mining.
-
PCAR uses Parallel Coordinates to visualize Association Rules (AR) that shares the same consequent.
The dimensions of the graph are the items of the AR. Each rule crosses the axis according to the
utility provided by the corresponding item to the rule.
-
TimeSearcher 2 builds on previous work at the University of Maryland
that explored the use of Timeboxes to query time series data.
The tool allow a multivariate visualization of time series data sets and enable the user to explore them
by means of zooming, panning and search features.
-
dib3D is a project that offers a 3d visualization of the Dept. of Computer
Science of the University of Bari. The aim of this project is to provide an
alternative view of the curses web site of the department. The main services
offered
-
description
-
PHiP (Patient History in Pocket) is a tool designed
to support neurologists in the treatment of patients with epilepsy;
it is intended to make available on mobile devices some sections of patient records
and to implement features and functionalities that neurologists consider needful to treat
such neurological diseases.