Open AccessMachine-learning methods for text named entity recognition
Author(s) -
Олександр Марченко
Publication year - 2016
Publication title -
problemy programmirovaniâ
Language(s) - English
Resource type - Journals
ISSN - 1727-4907
DOI - 10.15407/pp2016.02-03.150
Subject(s) - conditional random field , named entity recognition , computer science , artificial intelligence , machine learning , natural language processing , naive bayes classifier , precision and recall , recall , bayes' theorem , pattern recognition (psychology) , bayesian probability , support vector machine , engineering , linguistics , philosophy , systems engineering , task (project management)
The article describes machine learning methods for the named entity recognition. To build named entity classifiers two basic models of machine learning, The Naїve Bayes and Conditional Random Fields, were used. A model for multi-classification of named entities using Error Correcting Output Codes was also researched. The paper describes a method for classifiers' training and the results of test experiments. Conditional Random Fields overcome other models in precision and recall evaluations.