|
|
Sentiment Analysis in Automotive Industry
Bezák, Adam ; Otrusina, Lubomír (referee) ; Smrž, Pavel (advisor)
The main theme of this thesis is to familiarize with the basic methods of sentiment analysis on social networks. Thesis’s theme is aimed on the automotive industry, although this prinicipal can be used in any different examined branch. The basis of the practical part is to obtain data from the social networks, analyze them and then index them into ElasticSearch database. Another goal of the thesis is to visualize these data by means of a web portal. Created web portal provides various statistics of the leading automobile brands, an overview of new trends or the aspect visualization of the individual cars.
|
|
|
Automatic dictionary acquisition from parallel corpora
Popelka, Jan ; Pecina, Pavel (advisor) ; Mareček, David (referee)
In this work, an extensible word-alignment framework is implemented from scratch. It is based on a discriminative method that combines a wide range of lexical association measures and other features and requires a small amount of manually word-aligned data to optimize parameters of the model. The optimal alignment is found as minimum-weight edge cover, selected suboptimal alignments are used to estimate confidence of each alignment link. Feature combination is tuned in the course of many experiments with respect to the results of evaluation. The evaluation results are compared to GIZA++. The best trained model is used to word-align a large Czech-English parallel corpus and from the links of highest confidence a bilingual lexicon is extracted. Single-word translation equivalents are sorted by their significance. Lexicons of different sizes are extracted by taking top N translations. Precision of the lexicons is evaluated automatically and also manually by judging random samples.
|
|
|
News Feed Classifications to Improve Volatility Predictions
Pogodina, Ksenia ; Šopov, Boril (advisor) ; Červinka, Michal (referee)
This thesis analyzes various text classification techniques in order to assess whether the knowledge of published news articles about selected companies can improve its' stock return volatility modelling and forecasting. We examine the content of the textual news releases and derive the news sentiment (po larity and strength) employing three different approaches: supervised machine learning Naive Bayes algorithm, lexicon-based as a representative of linguistic approach and hybrid Naive Bayes. In hybrid Naive Bayes we consider only the words contained in the specific lexicon rather than whole set of words from the article. For the lexicon-based approach we used independently two lexicons one with binary another with multiclass labels. The training set for the Naive Bayes was labeled by the author. When comparing the classifiers from the machine learning approach we can conclude that all of them performed similarly with a slight advantage of the hybrid Naive Bayes combined with multiclass lexicon. The resulting quantitative data in form of sentiment scores will be then incorpo rated into GARCH volatility modelling. The findings suggest that information contained in news feeds does bring an additional explanatory power to tradi tional GARCH model and is able to improve it's forecast. On the...
|
|
|
Comparison of a traditional dictionary description and a corpus of written Czech with regard to semantic prosody
Vovchuk, Oleksandr ; Cvrček, Václav (advisor) ; Hudousková, Andrea (referee)
Czech dictionaries were created in the pre-corpus era; it is thus clear that some of their entries don't take into account semantic prosody - the fact that some lexemes occur in particular contexts (consequences are always far-reaching or catastrophic, while intention can be both evil and noble). The aim of this thesis is to compare selected parts of a dictionary (randomly selected probe into Czech adjectives) with corpus material, define the extent of missing information in the current state of Czech language description and explore, how many per cent of information about entries is missing in contemporary dictionaries. We based our research on Slovník spisovného jazyka českého (Dictionary of Written Czech) and representative corpuses of written language SYN2005, or else SYN2010. Context analysis was carried out by means of statistical methods and collocation rates. The difference between dictionary definitions and information inferred from the corpus research could become a further guideline for creating a new dictionary (besides adding a whole range of new entries that are still missing in Czech dictionaries). Keywords lexicon, corpus, semantic prosody
|
|
|
Vocabulary of a Game Mölkky
Červ, Petr ; Janovec, Ladislav (advisor) ; Vlčková, Jana (referee)
This bachelor thesis deals with the vocabulary of a game mölkky. The first part defines the basic terms and definitions. The next part describes the game of mölkky, its rules, history and present. The main part of the thesis is a selection of the most used words and phrases in the field of mölkky. Selected terms are analyzed in terms of content and form.
|
|
| |
|
|
News Feed Classifications to Improve Volatility Predictions
Pogodina, Ksenia ; Šopov, Boril (advisor) ; Červinka, Michal (referee)
This thesis analyzes various text classification techniques in order to assess whether the knowledge of published news articles about selected companies can improve its' stock return volatility modelling and forecasting. We examine the content of the textual news releases and derive the news sentiment (po larity and strength) employing three different approaches: supervised machine learning Naive Bayes algorithm, lexicon-based as a representative of linguistic approach and hybrid Naive Bayes. In hybrid Naive Bayes we consider only the words contained in the specific lexicon rather than whole set of words from the article. For the lexicon-based approach we used independently two lexicons one with binary another with multiclass labels. The training set for the Naive Bayes was labeled by the author. When comparing the classifiers from the machine learning approach we can conclude that all of them performed similarly with a slight advantage of the hybrid Naive Bayes combined with multiclass lexicon. The resulting quantitative data in form of sentiment scores will be then incorpo rated into GARCH volatility modelling. The findings suggest that information contained in news feeds does bring an additional explanatory power to tradi tional GARCH model and is able to improve it's forecast. On the...
|
|
|
Sentiment Analysis in Automotive Industry
Bezák, Adam ; Otrusina, Lubomír (referee) ; Smrž, Pavel (advisor)
The main theme of this thesis is to familiarize with the basic methods of sentiment analysis on social networks. Thesis’s theme is aimed on the automotive industry, although this prinicipal can be used in any different examined branch. The basis of the practical part is to obtain data from the social networks, analyze them and then index them into ElasticSearch database. Another goal of the thesis is to visualize these data by means of a web portal. Created web portal provides various statistics of the leading automobile brands, an overview of new trends or the aspect visualization of the individual cars.
|
|
|
Adjectival networks. On the grammar of French denominal adjectives
Strnadová, Jana ; Štichauer, Pavel (advisor) ; Radimský, Jan (referee) ; Namer, Fiammetta (referee)
in English This dissertation studies su xal derivation of adjectives from nouns in French. It is based on a lexicon of about 15, 000 adjectives, 40% of which may be considered deno- minal. I rst present the data under investigation. I describe the Dénom database, which was derived from large scale lexica. In order to assess the position of denominal adjectives in the more general adjectival system, I present a classi cation of French adjectives on the basis of their morphological properties. In the process, I spot cases where the fringes of the class of denominals are unclear, and question the distributional and semantic co- hesion of the class. I nally review di erent types of formal or semantic mismatches between the adjective and its base noun. In a second step, I present a study of the formal and semantic properties of a subset of denominal adjectives where the morphological relation between base and derivative is regular. This subset is selected on the basis of the type frequency of formal patterns of alternation between base and derivative. I describe the phonological and morpholo- gical properties of base nouns, with the aim of uncovering factors that play a role in the formation of adjectives. This leads to the observation of morphological niches, that is, cases where the presence of a...
|
|
|
Language transformation in TV news in 1985 and 2010
Geršáková, Jana ; Čmejrková, Světla (advisor) ; Gebhartová, Markéta (referee)
This thesis consists of analysis and mutual comparison of television newscast Události and Televizní noviny. Thesis analyses language modifications and selection of language resources in analogue programmes - with the broadcast difference of 25 years on public television. The work presents lexical, syntactic and phonological analysis. Language usage is also represented by a frequency dictionary, which is used in newscast.
|
guest ::
