This thesis is focused on the problem of semantic similarity of words in English language. At first reader is informed about theory of word sense clustering, then there are described chosen methods and tools related to the topic. In the practical part we design and implement system for determining semantic similarity using Word2Vec tool, particularly we focus on biomedical texts of MEDLINE database. At the end of the thesis we discuss reached results and give some ideas to improve the system.

This master's thesis describes creation of recommender system that is used in real server cbdb.cz. A~fully operational recommender system was developed using collaborative and content-based filtering techniques. Thanks to many user feedback, we were able to evaluate their opinion. Many recommended books were tagged as desirable. This thesis is extending current functionality of cbdb.cz with recommender system. This system uses its extensive database of ratings, users and books.

This thesis deals with problems of recommender systems and their usage in web applications. There are three main data mining techniques summarized and individual approaches for recommendation. Main part of this thesis is a suggestion and an implementation of web applications for recommending dishes from restaurants. Algorithm for food recommending is designed and implemented in this paper. The algorithm deals with the problem of frequently changing items. The algorithm utilizes hybrid filtering technique which is based on content and knowledge. This filtering technique uses cosine vector similarity for computation.

This thesis deals with description of existing methods of image retrieval. It contains set of methods for image description, coding of global and local descriptor (SIFT, etc.) and describes method of effective searching in multidimensional space (LSH). It continues with proposal and testing of three global descriptors using color histograms, histogram of gradients and the combination of both. The last part deals with similar image retrieval using proposed descriptors and the indexing method LSH and compares the results with the existing method. Product of this work is an experimental application which demonstrates the proposed solution.

The work deals with the design of a system foron-line analysis of web page similarity. The system consists of two main parts: A server part, which is based on Java EE and a client part, which works as an extension of a web browser. The master's thesis also solves the communication between these two components and their implementation. There are described two main approaches to the comparison that focus on visual features of the web page. The first approach is based on an analysis of the rendered image of the web page and the second one is based on the source code analysis.

This thesis deals with automatic acquiring thesauri from Wikipedia. It describes Wikipedia as a suitable data set for thesauri acquiring and also methods for computing semantic similarity of terms are described. The thesis also contains a description of concepts and implementation of the system for automatic thesauri acquiring. Finally, the implemented system is evaluated by the standard metrics, such as precision or recall.

Obsahem této práce je návrh metody porovnání webových stránek na základě vizuálních rysů. Na začátku jsou popsány možné přístupy k porovnávání dokumentů s ohledem na jejich použití. Dále jsou prezentovány přístupy porovnávání zaměřené na vizuální vzhled dokumentů. Zde jsou nejdříve popsány metody pro porovnávání z vyrendrovaného obrázku dokumentu a pak přístup pomocí zdrojového kódu. Tato práce je dále detailně zaměřena na získání vizuální rysů ze zdrojového kódu dokumentu. Je zde popsán návrh nové metody pro porovnávání dokumentů na základě vizuálních rysů, která využívá strukturální popis dokumentu. Součástí je taky popis implementace aplikace a dosažené výsledky. V závěru jsou informace o možném rozšíření navržené metody a dalším pokračování.

There are billions of photos on the internet and as the size of these digital repositories grows, finding target picture becomes more and more difficult. To increase the informational quality of photo albums we propose a new method that selects representative pictures from a group of photographs using computer vision algorithms. The aim of this study is to analyze the issues about image features, image similarity, object clustering and examine the specific characteristics of photographs. Tests show that there is no universal image descriptor that can easily simulate the process of clustering performed by human vision. The thesis proposes a hybrid algorithm that combines the advantages of selected features together using a specialized multiple-step clustering algorithm. The key idea of the process is that the frequently photographed objects are more likely to be representative. Thus, with a random selection from the largest photo clusters certain representative photos are obtained. This selection is further enhanced on the basis of optimization, where photos with better photographic properties are being preferred.

The aim of this work is to create a web application for automatic topic detection and segmentation of on-line courses. During playback of processed records, the application should be able to offer records from thematically consistent on-line courses. This document contains problem description, list of used instruments, description of implementation, the principle of operation and description of final user interface.

Cette thèse s'inscrit dans le cadre de la biologie des systèmes et porte plus particulièrement sur un problème relatif aux réseaux biologiques hétérogènes. Elle se concentre sur les relations entre le métabolisme et le contexte génomique, en utilisant une approche de fouille de graphes.Il est communément admis que des étapes enzymatiques successives impliquant des produits de gènes situés à proximité sur le chromosome traduisent un avantage évolutif du maintien de cette relation de voisinage au niveau métabolique ainsi que génomique. En conséquence, nous choisissons de nous concentrer sur la détection de réactions voisines catalysées par des produits de gènes voisins, où la notion de voisinage peut être modulée en autorisant que certaines réactions et/ou gènes soient omis. Plus spécifiquement, les motifs recherchés sont des trails de réactions (c'est-à-dire des séquences de réactions pouvant répéter des réactions, mais pas les liens entre elles) catalysées par des produits de gènes voisins. De tels motifs de voisinage sont appelés des motifs métaboliques et génomiques.De plus, on s'intéresse aux motifs de voisinage métabolique et génomique conservés, c'est-à-dire à des motifs similaires pour plusieurs espèces. Parmi les variations considérées pour un motif conservé, on considère l'absence/présence de réactions et/ou de gènes, ou leur ordre différent.Dans un premier temps, nous proposons des algorithmes et des méthodes afin d'identifier des motifs de voisinage métabolique et génomique conservés. Ces méthodes sont implémentées dans le pipeline libre CoMetGeNe (COnserved METabolic and GEnomic NEighborhoods). À l'aide de CoMetGeNe, on analyse une sélection de 50 espèces bactériennes, en utilisant des données issues de la base de connaissances KEGG.Dans un second temps, un développement de la détection de motifs conservés est exploré en prenant en compte la similarité chimique entre réactions. Il permet de mettre en évidence une classe de modules métaboliques conservés, caractérisée par le voisinage des gènes intervenants. / This thesis fits within the field of systems biology and addresses a problem related to heterogeneous biological networks. It focuses on the relationship between metabolism and genomic context through a graph mining approach.It is well-known that succeeding enzymatic steps involving products of genes in close proximity on the chromosome translate an evolutionary advantage in maintaining this neighborhood relationship at both the metabolic and genomic levels. We therefore choose to focus on the detection of neighboring reactions being catalyzed by products of neighboring genes, where the notion of neighborhood may be modulated by allowing the omission of several reactions and/or genes. More specifically, the sought motifs are trails of reactions (meaning reaction sequences in which reactions may be repeated, but not the links between them). Such neighborhood motifs are referred to as metabolic and genomic patterns.In addition, we are also interested in detecting conserved metabolic and genomic patterns, meaning similar patterns across multiple species. Among the possible variations for a conserved pattern, the presence/absence of reactions and/or genes may be considered, or the different order of reactions and/or genes.A first development proposes algorithms and methods for the identification of conserved metabolic and genomic patterns. These methods are implemented in an open-source pipeline called CoMetGeNe (COnserved METabolic and GEnomic NEighborhoods). By means of this pipeline, we analyze a data set of 50 bacterial species, using data extracted from the KEGG knowledge base.A second development explores the detection of conserved patterns by taking into account the chemical similarity between reactions. This allows for the detection of a class of conserved metabolic modules in which neighboring genes are involved.