The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
As razões de verossimilhança (LRs), que são comumente usadas para processamento de dados probabilísticos, são frequentemente estimadas com base nas contagens de frequência de elementos individuais obtidos de amostras. No processamento de linguagem natural, um elemento pode ser uma sequência contínua de N itens, chamados de N-grama, em que cada item é uma palavra, letra, etc. Neste artigo, tentamos estimar LRs com base em Ninformações de frequência de -grama. Uma abordagem de estimativa ingênua que usa apenas Nfrequências de -grama são sensíveis a baixas frequências (raras) N-gramas e não aplicável à frequência zero (não observado) N-gramas; estes são conhecidos como problemas de frequência baixa e zero, respectivamente. Para resolver esses problemas, propomos um método para decompor N-gramas em unidades de itens e, em seguida, aplicando suas frequências junto com o original Nfrequências de -grama. Nosso método pode obter as estimativas de N-gramas usando as frequências unitárias. Embora o uso apenas de frequências unitárias ignore as dependências entre os itens, nosso método aproveita o fato de que certos itens muitas vezes co-ocorrem na prática e, portanto, mantém suas dependências usando o relevante Nfrequências de -grama. Também introduzimos uma regularização para obter uma estimativa robusta para casos raros. N-gramas. Nossos resultados experimentais demonstram que nosso método é eficaz na resolução de ambos os problemas e pode controlar efetivamente as dependências.
Masato KIKUCHI
Nagoya Institute of Technology
Kento KAWAKAMI
LINE Corporation
Kazuho WATANABE
Toyohashi University of Technology
Mitsuo YOSHIDA
Toyohashi University of Technology
Kyoji UMEMURA
Toyohashi University of Technology
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Masato KIKUCHI, Kento KAWAKAMI, Kazuho WATANABE, Mitsuo YOSHIDA, Kyoji UMEMURA, "Unified Likelihood Ratio Estimation for High- to Zero-Frequency N-Grams" in IEICE TRANSACTIONS on Fundamentals,
vol. E104-A, no. 8, pp. 1059-1074, August 2021, doi: 10.1587/transfun.2020EAP1088.
Abstract: Likelihood ratios (LRs), which are commonly used for probabilistic data processing, are often estimated based on the frequency counts of individual elements obtained from samples. In natural language processing, an element can be a continuous sequence of N items, called an N-gram, in which each item is a word, letter, etc. In this paper, we attempt to estimate LRs based on N-gram frequency information. A naive estimation approach that uses only N-gram frequencies is sensitive to low-frequency (rare) N-grams and not applicable to zero-frequency (unobserved) N-grams; these are known as the low- and zero-frequency problems, respectively. To address these problems, we propose a method for decomposing N-grams into item units and then applying their frequencies along with the original N-gram frequencies. Our method can obtain the estimates of unobserved N-grams by using the unit frequencies. Although using only unit frequencies ignores dependencies between items, our method takes advantage of the fact that certain items often co-occur in practice and therefore maintains their dependencies by using the relevant N-gram frequencies. We also introduce a regularization to achieve robust estimation for rare N-grams. Our experimental results demonstrate that our method is effective at solving both problems and can effectively control dependencies.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2020EAP1088/_p
Copiar
@ARTICLE{e104-a_8_1059,
author={Masato KIKUCHI, Kento KAWAKAMI, Kazuho WATANABE, Mitsuo YOSHIDA, Kyoji UMEMURA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Unified Likelihood Ratio Estimation for High- to Zero-Frequency N-Grams},
year={2021},
volume={E104-A},
number={8},
pages={1059-1074},
abstract={Likelihood ratios (LRs), which are commonly used for probabilistic data processing, are often estimated based on the frequency counts of individual elements obtained from samples. In natural language processing, an element can be a continuous sequence of N items, called an N-gram, in which each item is a word, letter, etc. In this paper, we attempt to estimate LRs based on N-gram frequency information. A naive estimation approach that uses only N-gram frequencies is sensitive to low-frequency (rare) N-grams and not applicable to zero-frequency (unobserved) N-grams; these are known as the low- and zero-frequency problems, respectively. To address these problems, we propose a method for decomposing N-grams into item units and then applying their frequencies along with the original N-gram frequencies. Our method can obtain the estimates of unobserved N-grams by using the unit frequencies. Although using only unit frequencies ignores dependencies between items, our method takes advantage of the fact that certain items often co-occur in practice and therefore maintains their dependencies by using the relevant N-gram frequencies. We also introduce a regularization to achieve robust estimation for rare N-grams. Our experimental results demonstrate that our method is effective at solving both problems and can effectively control dependencies.},
keywords={},
doi={10.1587/transfun.2020EAP1088},
ISSN={1745-1337},
month={August},}
Copiar
TY - JOUR
TI - Unified Likelihood Ratio Estimation for High- to Zero-Frequency N-Grams
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1059
EP - 1074
AU - Masato KIKUCHI
AU - Kento KAWAKAMI
AU - Kazuho WATANABE
AU - Mitsuo YOSHIDA
AU - Kyoji UMEMURA
PY - 2021
DO - 10.1587/transfun.2020EAP1088
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E104-A
IS - 8
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - August 2021
AB - Likelihood ratios (LRs), which are commonly used for probabilistic data processing, are often estimated based on the frequency counts of individual elements obtained from samples. In natural language processing, an element can be a continuous sequence of N items, called an N-gram, in which each item is a word, letter, etc. In this paper, we attempt to estimate LRs based on N-gram frequency information. A naive estimation approach that uses only N-gram frequencies is sensitive to low-frequency (rare) N-grams and not applicable to zero-frequency (unobserved) N-grams; these are known as the low- and zero-frequency problems, respectively. To address these problems, we propose a method for decomposing N-grams into item units and then applying their frequencies along with the original N-gram frequencies. Our method can obtain the estimates of unobserved N-grams by using the unit frequencies. Although using only unit frequencies ignores dependencies between items, our method takes advantage of the fact that certain items often co-occur in practice and therefore maintains their dependencies by using the relevant N-gram frequencies. We also introduce a regularization to achieve robust estimation for rare N-grams. Our experimental results demonstrate that our method is effective at solving both problems and can effectively control dependencies.
ER -