magistrsko delo
Abstract
Z razvojem področja globokega učenja, ki temelji na umetnih nevronskih mrežah, se danes poskušajo rešiti že znani problemi področja obdelave naravnega jezika. V tem magistrskem delu obravnavamo problem razpoznavanja in klasifikacije imenskih entitet z uporabo metod globokega učenja. V magistrski nalogi smo uporabili programski jezik Python in odprtokodno knjižnico Keras. Preizkusili smo različne arhitekture rekurentnih nevronskih mrež, ki uporabljajo pomnilne celice LSTM in GRU. Prav tako smo opravili različne poskuse, v katerih smo iskali optimalne parametre nevronske mreže z namenom natančnega razpoznavanja in klasifikacije imenskih entitet. Učenje nevronske mreže in vrednotenje modelov smo izvedli na korpusih, ki so bili predstavljeni na konferenci CONLL leta 2003.
Keywords
obdelovanje naravnega jezika;razpoznavanje imenskih entitet;umetne nevronske mreže;diplomske naloge;
Data
Language: |
Slovenian |
Year of publishing: |
2019 |
Typology: |
2.09 - Master's Thesis |
Organization: |
UM FERI - Faculty of Electrical Engineering and Computer Science |
Publisher: |
L. Bašek |
UDC: |
004.032.26(043.2) |
COBISS: |
22167318
|
Views: |
1113 |
Downloads: |
172 |
Average score: |
0 (0 votes) |
Metadata: |
|
Other data
Secondary language: |
English |
Secondary title: |
Named Entity Recognition and Classification using Artificial Neural Network |
Secondary abstract: |
Deep learning growth based on artificial neural networks allowed us to solve well-known problems in the natural language processing field. In this Master's thesis we deal with the problem of identifying and classifying named entities using deep learning methods. In the project, we used the Python programming language and the Keras library. We tested different architectures of recurrent neural networks that use LSTM and GRU memory cells. We also performed various experiments in which we searched for the optimal parameters of the neural network with the intent to accurately recognize and classify name entities. Neural network learning and model evaluation were conducted at the corpora presented at the CONLL conference in 2003. |
Secondary keywords: |
natural language processing;named entity recognition;artificial neural networks;LSTM;GRU; |
URN: |
URN:SI:UM: |
Type (COBISS): |
Master's thesis/paper |
Thesis comment: |
Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije |
Pages: |
XIV, 90 f. |
ID: |
11001986 |