Ivan Jelínek

Benchmarking RDF production tools

Since a big part of web content is stored in relational databases (RDB) there are several approaches for generating of semantic web metadata from RDB. In our previous work we designed a novel approach for RDB to RDF data transformation. This paper describes experimental comparison of our system with several means of RDF production. We benchmarked both systems for theRDBtoRDFtransformation and native RDF repositories. The test results show a good performance of our system but also bring a new look at the effectiveness of the RDF production.

Using AICC to create reusable adaptive hypermedia e-learning content

Adaptive hypermedia systems represent one of promising future Internet concepts, especially in the area of e-learning. One of important contemporary claims is to create reusable adaptive hypermedia content, compatible with existing systems, to ease a widespread of this promising technology. In this paper we propose method of adaptive hypermedia e-learning courses construction using AICC communication standard. This proposal is based on formal description of adaptive hypermedia course topology and its mapping to AICC communication structure

Ontology-based approach to adaptation

The paper shows actual state of Adaptation techniques for use with the Semantic web and presents intended development. Focus is put on several areas: user profile acquisition and user profile sharing as step to cost-effective approach, ontology merging and support for natural language generation. Our intention is to point out general architecture for adaptation which can be used within the Semantic web.

Improvements of webometrics by using sentiment analysis for better accessibility of the web

The paper discusses the webometric model for effective acquirement of relevant information from the web that would better separate the useful data from the useless. Our research emphasis has been placed on techniques that would better reflect the semantic content of single pages. Webometrics is purely a quantitative approach to the web, which can be enhanced by qualitative methods and thereby allows us to expand the possibilities of a study problem. Sentiment analysis may be used as a qualitative complement to quantitative approach. This analysis provides a technique of sophisticated analysis of sentences using mathematical and statistical methods and linguistic analysis of text. Extension of the webometric techniques of sentiment analysis methods leads up to a better machine understanding of a web page and its overall semantic meaning. It can be assumed that the designed model will reduce the irrelevant web search results and thereby facilitate user access to the information on the web. The introductory part of the paper explains the concept of the sentiment analysis and the basic functional background of the webometric techniques.

Data extraction by visual matching

This article deals with the problem of data extraction on the basis of visual appearance and the similarity of visual representation. This approach extends the existing approaches to the extraction of data on another dimension. In addition to the structure of data, their visual representation is taken into account. This article presents a new method of comparing the structure and content of a document using transformation of data visual representation. It is possible to search and extract the necessary data on the basis of similarity of the results of these transformations.

Using centralized element in P2P network for better community management

In this paper we are focusing on description of a new possible Internet searching system. It uses Peer-to-Peer (P2P) model to build a clustered logical network using a multi-ring topology. We are also aware of the fact that without centralized element is the searching very difficult. For that we introduce central data-topics managing server as the intermediary between user (P2P node) and desired information. When using both models - P2P and Client/Server - we introduce two-way searching approach. For better node identification we use the meta-addressing which is also necessary for managing node profiles.

Exploratory testing supported by automated reengineering of model of the system under test

Exploratory Testing technique is well applicable to software development projects, where test basis is not available (or at least not complete and consistent to the extent allowing the creation of efficient test cases). The key factor for the efficiency of this technique is a structured process for the recording of explored path in the system under test. This approach also allows the creation of the test cases during exploratory testing process. These test cases can be used in the following re-testing of the system. If performed manually, the efficiency of such process strongly depends on the team organization and systematic work of the individuals in the team. This process can be aided by an automated support. In the paper, a framework, which automatically records selected tester’s actions in the system under test is presented. From these recordings, a model of the screen and action flows is reengineered and test cases are prepared. Tester is also able to define more meta-data in the test cases during this process. The recorded model and defined test cases are then available for the next rounds of testing. The performed case study shows that Exploratory Testing aided by this machine support is less resource demanding than Exploratory Testing performed manually only. Also, larger part of SUT was explored during the tests, when this systematic support was available to testers.

Transformation of IFML schemas to automated tests

Interaction Flow Modelling Language (IFML) is a recently standardized modelling language designed to capture the structure and properties of the application user interface. In the paper we discuss suitability of the IFML model for automatic generation of automated application front--end tests, as detailed model of the application front--end provides enough information to exercise particular actions in the tested application front--end to verify expected behaviour. Our approach is based on the multi--level transformation of an IFML model to automated tests. The IFML model is converted to an application front--end test model. Then abstract test cases are created from this model using set of defined rules. Finally the abstract test cases are transformed to a particular physical automated test cases using a template engine.

Modelling of the musical taste using semantic network for recommendation systems

A rapid growth of the Internet and the multimedia industry has created the current need for effective information filtering systems and, in particular, recommendation systems. The main purpose of recommendation systems in music field is to provide interesting and unknown music artists to the end-user, being based on users musical taste. Modelling of musical taste is the key step of developing recommendation system. In this paper we suggest a method of modelling users musical taste based on undirected semantic network. To enrich the model of users musical taste we use information from public databases. This method relies on Social Network Analysis and Betweenness centrality measure to identify factors which influence the users choice.

Proposing of modular system for web information extraction

The paper discusses dimensions and parameters of web information extraction such as level of automatism, type of source document related to its structuring and dependence on extractable domain. In second half we analyze extraction process in more detail according to considered dimensions, propose its phases and discuss modules for each one and interfaces between phases.