Background:

State based specifications are used to define the aspects of a system that are (later on) to be implemented. However, studies indicate that specifications are not so easy to read, to comprehend, and, thus, that they are not easy to maintain. Though there are tools for dealing with larger specifications, comprehension is impeded due to their performance limitations. Dependencies have to be re-calculated in a time-consuming process. Storing the specification and meta-information in a reusable form would improve the situation for maintenance activities.

Aim:

The objective of this work is to identify and define a database schema for depicting formal specifications, to map Z specifications to a database representation, and, finally, to define operations for adding/deleting/modifying parts of the specification.

Background:

Studies indicate that formal specifications are not so easy to read, to comprehend, and, thus, that they are not easy to maintain. The reason is that formal specifications can be seen as complex structures that contain basic specification primes interlinked by various dependencies. This structure can best be expressed by the visualization as a graph. Taking a closer look at the graph yields substructures, regions that are more closely related to each other. Partitioning the graph in respect to these regions gives hints towards the structure and improves comprehension activities.

Aim:

The objective of this work is to identify and compare graph clustering algorithms (Markov Cluster, Iterative Conductance Cutting, Geometric MST Clustering) in respect to their suitability to the graph representation of formal Z specifications, and finally to implement a library providing the algorithms in Java.

Successful SW-projects stand out for conscientious planning, tracking, and post-factum analysis. Typically, these steps are also supported by a software tool, a so-called PM-System. Carinthian SMEs’ requirements for such a tool differ in some aspects from the general requirements. This work should define the basis for a PM software for SMEs.

Aim:

The objectives of this work are as follows: identification of components of PM-Systems for a CMMI-Level 2 organization, calibration of requirements, conducing a survey in order to gain an overview of existing systems, amd finally, creation of an open architecture for a PM-System that also allows for higher CMM levels.

The task is to determine whether a spreadsheet program becomes too complex to maintain (and should be substituted by other software systems) or not. This means also to provide answers to the common question “How complex is too complex”. In analogy to SW systems common heuristics and metrics will be needed, thus after an initial discussion of the appropriateness of such approaches, some spreadsheet specific metrics and their theoretical underpinnings (graph theory, spreadsheet structures) as well as future metrics should be analysed.

Aim:

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. The basis for the simulation run is a rule-based system which, for different simulation tasks, contains up to 500 rules. For assessing simulation runs these rules are of high interest, as they define the behaviour of the system under specific situations. Thus an assessment component explaining the effect of a decision has to be aware of relevant rules. At the moment these rules and their explanation texts are defined by hand.

Aim:

The objective of this work is to define methods that (semi-)automatically derive rules and effects from the basic rule-set, and thus provides the basis for an automatic assessment component.

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. For assessment all decisions and effects are stored in a database. The game trace is displayed on a perspective wall and areas of interest (decisions which were critical or excellent) are hard to identify.

Aim:

The objectives of this work are to define rules for the identification of regions of interest, and to extend the concept of a perspective wall (icon based) in order to display these regions in the game trace.

State based specifications are used to define the aspects of a system that is (later on) to be implemented. However, studies indicate that specifications are not so easy to read, to comprehend, and, thus, that easy to maintain. Furthermore they do also contain errors.

Aim:

On the first sight formal Z specifications seem to be roughly unstructured text. However, when taking a closer look at them dependencies between parts of the specification can be identified. These dependencies are calculated by looking at identifiers that are defined and/or used in the text.

Aim:

This task aims at applying immune systems for Text Filtering (TF). In the general framework of information retrieval, TF, as its name implies, is a task that selects only documents that are of interest to a specific user. TF systems aim at supporting users in finding relevant information from a dynamic stream of documents like Internet. TF systems can be used by Internet search engines, by anti-spam agents, etc. The crucial issue in any information filtering system is the generation of the user's profile. A profile is a set of terms that describe the likes (and probably dislikes) of the user. Given a stream of documents, then whenever a relevant document from that stream is found, it is automatically sent to the user.

Aim:

Rule learning seeks to infer knowledge in the form of IF-THEN rules from a set of data examples. Relying on clustering, fuzzy classification rules can be discovered from data by building some clusters that facilitate the interpretation of the knowledge hidden within data.

Aim:

Classification with partially labelled data involves learning from a few labelled examples as well as a large number of unlabelled examples, and represents a combination of supervised and unsupervised learning resulting in what is known as semi-supervised learning. Unlabelled examples provide descriptive information about the input domain distribution, but only the labelled examples indicate the actual classification that underlies the data.

Aim:

On the first sight formal Z specifications seem to be roughly unstructured text. However, when taking a closer look at them, higher-level structures can be identified. These structures are based on the syntactical layout and on hidden dependencies. In the literature one can also find some approaches that transform specifications to UML.

Aim:

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. All data is stored in a MySQL DB using JDBC as a communication interface layer. Experiments, though, have shown that EJB is a valid alternative to JDBC.

Aim:

There are many clone detection tools based on either variants of pattern matching algorithms or AST comparisons, each of them with their own pros and cons.

Aim:

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. The idea behind AMEISE is also to sustain the trainee during a simulation run. Therefore agents (and tutors) store additional data into the AMEISE SQL database. However, up to now there is no component identifying decision-chains and ripple-effects.

Aim:

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. The assessment during and at the end of a simulation run is of high interest and is partially supported by the system (as every decision and step is stored in the database).

Aim:

The objective of this work is to implement a component that is able to generate configurable assessment reports as needed during AMEISE courses.

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. For assessment all decisions and effects are stored in a database.

Aim:

The objective of the work is to explore the expressiveness of a time wall for displaying an AMEISE game trace (which means the log of decisions and effects).

AMEISE (http://ameise.uni-klu.ac.at/) is a system for the simulation of (virtual) SW projects. The basic idea is to enable a trainee to head (as a project manager) a middle-sized SW project and to let him/her experiment with project management decisions. Up to now the simulation trace is represented as a long, unstructured text.

Aim:

Digital libraries are becoming great sources of knowledge. Given the huge number of XML documents stored in these libraries, it is too costly to search for a particular document. An optimal way to reduce the access time to information is to organize the library in the form of classes where similar documents are assembled together.

Aim: