Ing. Martin Řimnáč, Ph.D.

The data structure estunatimation, given for an example from web sources, can be used not only for a data storage, but also for advanced data querying. The paper extends the data structure estimation method, giving the relational schema corresponding to an input data set, and describes the data storage by the easy-designed associative repository based on the estimated structure. Two implematations are presented: The first one saving data as instances of functional dependencies and the second one saving only instances of functional dependencies between single attributes extended by metainformation about complex attributes.

Mechanisms for accessing and training the data repositury using a binary matrix formalism are presented. The repository is designed for a data storage trought corresponding instance of single attribute functional dependencies, which can be seen as similar to the binary predicate formalism being used by RDF semantic web format.Two mechanisms for quering a repository, generalization and specialization, are given. Futhermore, the incremental repository training mechanism with no extra requirements on the input data form is described: The extensional functional dependency system is used s a generalized view on the storage data; the algorithm is inspired by the functional dependency discovery approach.

The data structure estimation is one of ways data can be automatically interpreted leading to a functional dependency system description. The model building can be compared with some machine learning methods.The paper summarizes basic inductive logic programming methods and analyses them in the context of the data structure estimation. Finally, the paper proposes that some basic task of ILP can be solved by data structure estimation method and gives an example.

The data structure estimation may provide a connection between the semantic web vision and curently used web sources containing no semantic description of presented information. It enables information from web pages to be processed by semantic web tools with the best to be known about these sources, with the source estimated data structure. The paper gives a method description and presents its usage for a inductive logic programming task. The advantages of a rule-based system usage is given as well.

The article introduces into automatic searching of web pages. It describes the problems that are encountered while crawling web pages and it explains the calculation of page rank valuation of the pages.

The article introduces into structure of searching engine Google and problem with SEO (search engine optimization).

The paper describes the database model reconstruction method, which generates from an input data set the hypotheses about a model. There are also included basic model properties and the method inference step by step.