AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


Text Categorization Using Hierarchical Bayesian Network Classifiers

Author

Term

10. Term

Publication year

2002

Abstract

Denne afhandling introducerer hierarkiske Bayesianske netværk (HBN) som klassifikatorer, en type probabilistisk model til automatisk at tildele emner til tekst. Bayesianske netværk repræsenterer relationer mellem variable som en rettet graf og muliggør slutning under usikkerhed. En hierarkisk struktur tilføjer niveauer, der kan fange generelle til mere specifikke mønstre i dokumenter. Vi præsenterer algoritmer til at konstruere HBN-klassifikatorer ud fra data og afprøver dem på Reuters' test-samling for tekstkategorisering.

This thesis introduces hierarchical Bayesian network (HBN) classifiers, a type of probabilistic model for automatically assigning topics to text. Bayesian networks represent relationships between variables as a directed graph and support reasoning under uncertainty. Adding a hierarchy introduces levels that can capture general-to-specific patterns in documents. We present algorithms for constructing HBN classifiers from data and test them on the Reuters text categorization test collection.

[This abstract was generated with the help of AI]