diplomsko delo
Sašo Hrnčić (Author), Tomaž Kosar (Mentor), Vili Podgorelec (Co-mentor)

Abstract

Cilj diplomske naloge je izdelati preprost kategorizacijski sistem, ki zna nov tekstovni dokument čim natančneje uvrstiti v naprej definirane kategorije. Ena izmed funkcionalnosti sistema je prepoznavanje jezika, ki je bilo testirano na podatkovnih korpusih dokumentov Wikipedije, Europarla in jezikovnih modelov projekta LibTextCat. Kategorizacijski sistem je bil razširjen še na prepoznavanje v naprej definiranih tematikah korpusa 20 Newsgroups in Reuters-21578. Za predstavitev dokumentov smo uporabili n-gramsko tehniko, ki smo jo kombinirali s selekcijskimi in statističnimi metodami. Dosežene rezultate smo analizirali ter dokumentirali. Podrobneje smo predstavili problematiko, lastne izkušnje, lastnosti uporabljenih metod ter obstoječe raziskave.

Keywords

tekstovno kategoriziranje;n-grami;strojno učenje;teorija informacij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: S. Hrnčić
UDC: 004.05:004.5(043.2)
COBISS: 19991318 Link will open in a new window
Views: 730
Downloads: 50
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: SUPERVISED TOPICS' DETECTION BASED ON FEATURE SELECTION AND STATISTICAL METHODS
Secondary abstract: The main goal of diploma work is to develop simple text classification system that is able to automatically classify a document into predefined categories as accurately as possible. One of the functionalities of the system is language detection that has been tested on documents of Wikipedia, Europarl and language models of project LibTextCat. Classification system has been expanded to identify predefine topics of the corpus 20 Newsgroups and Reuters-21578. For document presentation we used n-grams technique, which was combined with feature selection methods and statistical methods. The obtained results were analyzed and documented. We also present text classification problem, our experiences, features of used methods and some existing research.
Secondary keywords: text classification;n-grams;machine learning;information theory;
URN: URN:SI:UM:
Type (COBISS): Undergraduate thesis
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informatika
Pages: XV, 58 str.
ID: 9162164