Entry Naptali:2010:TDL from talip.bib
Last update: Sun Oct 15 02:55:04 MDT 2017
Top |
Symbols |
Numbers |
Math |
A |
B |
C |
D |
E |
F |
G |
H |
I |
J |
K |
L |
M |
N |
O |
P |
Q |
R |
S |
T |
U |
V |
W |
X |
Y |
Z
BibTeX entry
@Article{Naptali:2010:TDL,
author = "Welly Naptali and Masatoshi Tsuchiya and Seiichi
Nakagawa",
title = "Topic-Dependent Language Model with Voting on Noun
History",
journal = j-TALIP,
volume = "9",
number = "2",
pages = "7:1--7:??",
month = jun,
year = "2010",
CODEN = "????",
DOI = "https://doi.org/10.1145/1781134.1781137",
ISSN = "1530-0226 (print), 1558-3430 (electronic)",
ISSN-L = "1530-0226",
bibdate = "Mon Jun 21 18:03:02 MDT 2010",
bibsource = "http://portal.acm.org/;
http://www.math.utah.edu/pub/tex/bib/talip.bib",
abstract = "Language models (LMs) are an important field of study
in automatic speech recognition (ASR) systems. LM helps
acoustic models find the corresponding word sequence of
a given speech signal. Without it, ASR systems would
not understand the language and it would be hard to
find the correct word sequence. During the past few
years, researchers have tried to incorporate long-range
dependencies into statistical word-based $n$-gram LMs.
One of these long-range dependencies is topic. Unlike
words, topic is unobservable. Thus, it is required to
find the meanings behind the words to get into the
topic. This research is based on the belief that nouns
contain topic information. We propose a new approach
for a topic-dependent LM, where the topic is decided in
an unsupervised manner. Latent Semantic Analysis (LSA)
is employed to reveal hidden (latent) relations among
nouns in the context words. To decide the topic of an
event, a fixed size word history sequence (window) is
observed, and voting is then carried out based on noun
class occurrences weighted by a confidence measure.
Experiments were conducted on an English corpus and a
Japanese corpus: {\em The Wall Street Journal\/} corpus
and {\em Mainichi Shimbun\/} (Japanese newspaper)
corpus. The results show that our proposed method gives
better perplexity than the comparative baselines,
including a word-based/class-based $n$-gram LM, their
interpolated LM, a cache-based LM, a topic-dependent LM
based on $n$-gram, and a topic-dependent LM based on
Latent Dirichlet Allocation (LDA). The {\em n\/} -best
list rescoring was conducted to validate its
application in ASR systems.",
acknowledgement = ack-nhfb,
articleno = "7",
fjournal = "ACM Transactions on Asian Language Information
Processing",
journal-URL = "http://portal.acm.org/browse_dl.cfm?&idx=J820",
keywords = "Language model; latent semantic analysis; perplexity;
speech recognition; topic dependent",
}
Related entries
- $n$,
9(2)5,
9(3)11,
12(1)1,
12(4)15
- acoustic,
7(3)10,
10(1)6,
10(2)7
- allocation,
8(3)12
- analysis,
2(4)301,
3(2)94,
3(3)169,
4(3)263,
7(1)1,
7(1)3,
7(2)5,
7(3)9,
8(1)2,
8(3)12,
8(4)19,
9(2)6,
9(3)11,
9(4)15,
10(1)4,
10(3)16,
10(4)20,
11(2)4,
11(2)7,
11(4)16,
11(4)18,
12(2)6,
13(2)9,
13(3)11
- application,
1(2)103,
4(4)475,
6(2)7,
6(4)2,
7(1)2,
7(2)7,
7(3)8,
7(4)11,
7(4)13,
8(2)7,
8(2)9,
8(3)10,
8(4)14,
9(4)13,
10(2)10,
10(4)19,
11(4)17,
12(3)11
- ASR,
9(1)2,
10(2)7
- automatic,
1(3)207,
3(1)51,
5(2)165,
6(2)6,
7(1)2,
8(4)15,
8(4)16,
8(4)18,
9(1)2,
9(2)6,
10(1)5,
10(2)7,
10(3)12,
10(3)13,
10(4)21,
12(2)7,
12(4)14,
12(4)15,
13(3)13
- based,
1(3)269,
2(4)301,
3(2)113,
4(3)357,
4(4)377,
5(2)121,
5(2)165,
5(3)185,
6(2)8,
6(3)9,
6(3)11,
6(4)3,
7(2)6,
7(3)10,
7(4)12,
8(1)4,
8(3)11,
8(3)12,
8(4)17,
8(4)19,
9(1)2,
9(2)5,
9(2)6,
9(3)11,
10(1)4,
10(1)6,
10(2)7,
10(3)14,
10(3)16,
11(2)6,
11(3)8,
11(3)11,
11(4)14,
11(4)15,
11(4)16,
11(4)17,
11(4)18,
12(1)1,
12(1)2,
12(1)4,
12(3)9,
12(3)10,
12(3)11,
13(1)2,
13(1)3,
13(1)4,
13(2)10,
13(3)13,
13(4)17,
13(4)18
- based, word-,
1(3)173,
1(4)297,
6(3)9,
9(3)11,
11(2)7
- baseline,
7(1)1,
7(2)5,
7(2)6,
7(3)8,
8(1)4,
8(3)10,
9(1)3,
9(1)4,
9(2)6,
9(3)12,
11(4)13,
11(4)14
- best,
5(2)89,
5(3)183,
7(2)7,
7(4)13,
8(2)6,
9(1)3,
9(3)11,
9(3)12,
11(3)8,
11(4)13,
12(1)2,
12(2)5,
12(2)7,
12(3)9,
12(4)14,
13(1)4,
13(4)16
- better,
6(2)8,
7(2)7,
7(4)13,
8(1)4,
8(3)11,
8(4)16,
9(1)1,
9(3)11,
9(3)12,
11(3)8,
11(4)13,
11(4)15,
12(2)7
- carried,
8(4)18,
12(4)16
- class,
6(3)9,
10(1)2,
10(2)9,
10(3)14
- comparative,
4(2)159,
8(1)3,
9(3)11,
11(3)10
- conducted,
6(2)8,
7(4)12,
8(1)2,
9(3)10,
9(3)12,
11(2)4,
11(2)7,
13(1)3,
13(4)18
- confidence,
10(1)6,
10(2)7
- contain,
7(4)11,
8(3)11,
9(1)1,
10(4)21,
11(3)10,
12(4)16
- context,
6(4)2,
8(4)15,
9(4)14,
10(1)3,
10(4)18,
10(4)19,
11(2)6,
11(2)7,
11(3)9,
12(1)2,
13(1)2
- corpus,
4(1)18,
4(4)400,
5(2)165,
6(3)9,
6(3)11,
7(2)6,
7(3)9,
7(4)13,
8(4)17,
9(2)5,
9(4)14,
10(2)7,
11(2)5,
11(3)9,
11(3)10,
11(3)11,
11(4)16,
12(1)1,
12(1)2,
12(1)3,
13(1)3
- correct,
7(4)13,
10(2)10,
11(4)13,
12(3)9
- corresponding,
5(2)89,
7(3)10,
8(2)7,
9(3)12,
10(2)7,
11(2)5,
11(3)8
- decide,
11(4)14
- dependency,
6(3)9,
8(3)10,
10(3)15,
11(2)6,
11(4)18
- dependent,
13(4)17
- Dirichlet,
8(3)12,
11(2)4,
11(3)9
- during,
7(3)9,
11(1)1,
12(4)15,
13(3)14
- employed,
10(1)6,
11(2)5,
12(4)17
- English,
2(3)245,
4(2)135,
5(2)89,
5(2)121,
5(3)245,
6(2)6,
6(2)7,
6(3)11,
6(4)2,
7(1)1,
7(4)11,
8(2)9,
8(4)15,
8(4)16,
8(4)17,
9(1)1,
9(1)3,
9(3)12,
9(4)14,
9(4)15,
10(1)2,
10(1)4,
10(2)8,
10(3)14,
10(3)15,
10(4)17,
11(2)4,
11(2)5,
11(2)6,
11(3)8,
11(3)11,
12(2)5,
12(3)12,
12(4)14,
12(4)17,
13(1)1,
13(4)16
- event,
5(1)61,
6(3)11,
7(1)3,
9(4)15
- experiment,
2(2)101,
2(2)143,
5(2)146,
5(2)165,
5(3)245,
6(2)7,
6(2)8,
7(1)1,
7(1)2,
7(1)3,
7(2)5,
7(4)11,
8(1)2,
8(2)6,
8(3)11,
8(4)17,
9(2)6,
9(3)11,
9(3)12,
10(4)20,
11(2)4,
11(2)5,
11(2)7,
11(3)10,
11(3)11,
11(4)15,
11(4)17,
12(1)2,
12(2)5,
12(2)7,
12(3)9,
12(3)11,
12(3)12,
12(4)16,
12(4)17,
13(1)2,
13(1)3,
13(2)6,
13(2)8,
13(3)13,
13(4)18
- few,
8(2)9,
8(4)18,
9(3)12,
11(1)1,
13(2)6,
13(2)8
- field,
2(3)290,
6(4)3,
8(4)14,
9(2)5,
10(1)4,
10(1)6,
11(4)13,
12(1)2,
12(4)15
- find,
5(2)89,
6(4)2,
7(3)8,
9(3)11,
11(3)8,
11(3)9,
12(2)5,
13(3)12
- fixed,
9(3)12
- get,
8(4)15,
10(2)8
- give,
7(4)11,
11(2)7
- given,
5(2)89,
5(2)165,
7(3)10,
7(4)13,
8(1)2,
9(1)1,
11(2)5,
11(3)8,
11(3)11,
11(4)13,
12(4)14,
13(1)2,
13(4)17
- gram,
3(2)113,
6(2)6,
9(2)5,
9(3)11,
11(2)7,
12(1)1,
12(4)15
- hard,
6(2)8,
12(3)9
- have,
5(2)121,
6(3)10,
6(4)2,
7(2)7,
7(3)8,
7(4)11,
7(4)12,
7(4)13,
8(2)8,
8(2)9,
8(4)14,
8(4)16,
8(4)17,
8(4)18,
9(1)1,
9(1)3,
9(4)15,
10(1)5,
10(2)7,
10(4)19,
10(4)20,
10(4)21,
11(1)2,
11(2)6,
11(3)8,
12(1)3,
12(2)6,
13(1)3,
13(2)6,
13(2)7,
13(2)9,
13(3)11,
13(3)12
- help,
5(2)146,
6(2)6,
7(3)8,
8(4)17,
8(4)18,
9(1)3,
10(2)10,
11(3)9,
12(1)3,
12(4)15
- hidden,
3(2)146,
7(3)10,
8(2)6,
8(3)12,
10(4)20,
11(3)9,
13(3)12
- history,
8(1)2,
11(2)5,
11(4)13
- important,
5(4)413,
7(1)3,
7(3)10,
9(1)1,
9(4)13,
10(2)10,
11(4)18,
12(1)3,
13(3)12,
13(4)18
- including,
5(2)121,
6(2)7,
7(1)3,
7(2)7,
8(1)3,
8(2)6,
9(1)1,
9(3)11,
10(3)12,
11(4)18,
13(4)16
- incorporate,
8(2)9,
12(2)5,
12(3)12
- information,
1(1)65,
1(1)83,
1(4)281,
2(3)245,
2(3)295,
3(1)1,
3(4)227,
4(2)57,
4(2)78,
4(3)243,
4(3)357,
4(4)375,
4(4)475,
5(1)44,
5(2)89,
5(3)264,
5(4)291,
5(4)296,
5(4)323,
6(2)7,
6(4)2,
7(1)2,
7(1)3,
7(2)5,
7(3)8,
7(4)12,
8(1)2,
8(1)3,
8(3)10,
8(3)11,
8(4)15,
8(4)16,
8(4)17,
9(1)1,
9(3)9,
9(3)10,
9(3)11,
9(3)12,
9(4)13,
9(4)14,
9(4)15,
10(2)8,
10(2)10,
10(3)15,
10(3)16,
10(4)19,
10(4)20,
10(4)21,
11(1)2,
11(2)6,
11(2)7,
11(4)15,
11(4)18,
12(2)5,
12(3)11,
12(4)16,
13(2)7,
13(3)13
- Japanese,
4(3)263,
4(3)280,
4(4)375,
5(2)146,
5(3)264,
6(3)11,
6(4)1,
7(2)6,
7(2)7,
7(3)8,
8(2)9,
9(1)4,
11(2)6,
11(3)8,
11(3)11,
11(4)14,
11(4)18,
12(1)3,
12(2)7,
12(3)10,
12(3)11,
12(4)16,
12(4)17
- journal,
9(3)8,
13(3)15
- latent,
3(2)94,
7(1)3,
8(1)2,
8(3)12,
10(4)17,
11(3)11,
12(2)5
- list,
10(4)20,
11(3)11,
13(2)7
- LM,
8(1)2,
9(3)11
- manner,
8(3)12,
11(2)6,
12(3)9
- meaning,
3(1)33,
7(4)11,
9(1)3,
9(2)5,
11(4)14,
12(3)11,
12(4)16
- measure,
5(2)89,
6(2)6,
6(4)3,
8(2)7,
10(1)2,
10(1)6,
10(4)20,
11(2)6,
11(3)9,
11(3)11,
13(3)11,
13(3)13,
13(3)14
- n,
6(2)6,
11(2)7,
12(1)1
- new,
5(1)74,
5(2)121,
6(2)8,
6(3)9,
7(2)7,
7(4)11,
8(2)6,
9(1)4,
10(1)4,
10(3)15,
10(4)17,
10(4)20,
11(2)6,
12(1)1,
12(1)3,
12(3)8,
12(3)9,
12(4)14,
13(1)1,
13(1)2,
13(2)6
- newspaper,
4(3)321,
13(2)9
- not,
5(2)89,
5(2)121,
6(3)10,
7(2)7,
7(3)8,
7(3)10,
7(4)11,
7(4)12,
7(4)13,
8(1)3,
8(2)7,
8(2)8,
8(2)9,
8(4)17,
9(2)6,
9(4)14,
10(1)4,
10(1)5,
10(2)8,
10(2)10,
10(4)18,
11(1)2,
11(3)10,
12(1)2,
12(1)3,
12(2)7,
12(3)9,
13(2)7,
13(2)9,
13(4)17,
13(4)18
- noun,
3(1)33,
7(4)13,
9(1)1,
12(3)10
- observed,
8(1)2,
11(2)4
- one,
5(2)89,
5(2)121,
6(2)6,
6(3)9,
6(4)3,
7(3)8,
7(3)9,
7(4)11,
7(4)13,
8(2)9,
8(4)16,
8(4)17,
9(1)1,
9(2)5,
9(3)12,
9(4)14,
10(1)5,
10(3)12,
10(3)13,
10(4)19,
11(2)4,
11(2)6,
11(2)7,
11(4)14,
12(1)1,
12(1)2,
12(2)5,
12(2)7,
12(3)11,
12(4)16,
13(1)4,
13(2)10,
13(4)17,
13(4)18
- past,
7(4)11
- perplexity,
6(3)9,
12(4)14
- propose,
5(2)89,
6(2)6,
6(2)8,
6(3)11,
7(3)8,
7(3)10,
7(4)12,
8(1)2,
8(1)4,
8(2)9,
8(4)19,
9(4)13,
10(2)10,
10(3)12,
10(3)15,
10(4)17,
10(4)20,
11(2)7,
11(3)9,
11(3)11,
11(4)15,
11(4)16,
11(4)18,
12(1)1,
12(1)2,
12(1)3,
12(1)4,
12(2)5,
12(2)6,
12(3)9,
12(3)10,
12(3)12,
12(4)16,
13(1)2,
13(1)3,
13(2)8,
13(2)9,
13(3)12,
13(3)13,
13(4)17,
13(4)18
- proposed,
5(2)121,
5(2)165,
6(2)7,
7(1)1,
7(1)2,
7(3)9,
7(3)10,
7(4)11,
7(4)13,
8(1)4,
8(2)6,
8(3)10,
8(3)11,
8(4)14,
8(4)19,
9(1)1,
9(2)5,
10(2)7,
10(2)9,
10(3)14,
10(4)18,
11(1)3,
11(2)5,
11(2)6,
11(2)7,
11(3)8,
11(3)9,
11(3)10,
11(3)11,
11(4)16,
11(4)17,
12(1)4,
12(2)5,
12(2)7,
12(3)12,
12(4)16,
12(4)17,
13(2)6,
13(2)8,
13(3)13,
13(4)18
- recognition,
1(1)83,
1(4)297,
2(1)27,
2(3)290,
5(1)4,
5(2)165,
6(2)6,
6(3)9,
6(4)3,
7(1)2,
7(3)10,
8(1)2,
8(3)11,
8(4)18,
9(1)2,
10(1)6,
10(2)7,
10(2)9,
10(3)13,
11(1)1,
11(1)2,
11(4)13,
11(4)16,
11(4)17,
11(4)18,
12(1)4,
12(3)10,
13(3)12,
13(4)16
- relation,
1(3)173,
7(2)6,
7(4)12,
8(3)10,
10(3)14,
10(3)15,
11(1)3,
11(3)11,
11(4)16,
11(4)18,
12(1)3,
13(1)2,
13(1)3,
13(1)4
- rescoring,
8(1)4
- research,
2(2)101,
5(2)89,
7(4)12,
8(1)2,
8(1)4,
8(4)18,
9(3)12,
10(1)4,
10(3)14,
10(4)19,
10(4)20,
10(4)21,
11(1)1,
11(1)2,
11(3)10,
12(1)3,
12(4)14,
13(1)1,
13(2)7,
13(2)10
- researchers,
8(4)14,
11(1)1,
11(1)2,
11(3)8,
11(4)15,
13(1)3
- result,
4(2)135,
5(2)121,
5(2)146,
5(2)165,
6(2)6,
6(2)7,
6(3)9,
6(3)11,
6(4)3,
7(1)2,
7(2)5,
7(2)6,
7(2)7,
7(3)8,
7(3)10,
7(4)11,
7(4)12,
7(4)13,
8(1)2,
8(1)3,
8(1)4,
8(2)6,
8(2)9,
8(3)10,
8(3)12,
8(4)14,
8(4)15,
8(4)16,
8(4)17,
8(4)18,
8(4)19,
9(1)1,
9(1)2,
9(2)5,
9(2)6,
9(3)11,
9(3)12,
9(4)14,
10(1)2,
10(2)7,
11(2)4,
11(2)5,
11(3)8,
11(3)9,
11(3)11,
11(4)13,
11(4)14,
11(4)15,
12(1)3,
12(1)4,
12(2)5,
12(2)7,
12(3)9,
12(3)10,
12(3)11,
12(4)14,
12(4)16,
13(1)1,
13(1)4,
13(2)6,
13(2)7,
13(2)9,
13(3)11,
13(3)12,
13(3)14
- reveal,
9(3)11,
13(3)11,
13(3)13
- semantic,
3(1)66,
3(2)94,
5(2)165,
5(3)228,
7(1)2,
7(1)3,
7(4)13,
8(1)2,
8(2)6,
8(3)12,
8(4)15,
10(1)5,
10(3)13,
10(3)14,
10(3)15,
10(4)20,
10(4)21,
11(3)8,
11(3)11,
11(4)15,
11(4)18,
12(3)11,
12(4)16
- sequence,
5(2)165,
6(2)7,
6(3)10,
7(3)9,
8(1)2,
8(1)3,
9(3)12,
9(4)14,
10(1)4,
12(1)2,
12(1)3,
12(3)9,
12(3)10,
12(4)17,
13(1)2
- show,
5(2)89,
5(2)146,
7(1)1,
7(1)2,
7(1)3,
7(4)11,
7(4)12,
7(4)13,
8(1)4,
8(2)7,
8(2)9,
8(3)12,
8(4)16,
8(4)17,
9(1)1,
9(1)2,
9(1)3,
9(2)5,
9(2)6,
9(3)11,
9(3)12,
9(4)14,
10(1)3,
10(3)15,
11(2)4,
11(2)5,
11(2)7,
11(3)8,
11(3)11,
11(4)14,
11(4)15,
11(4)17,
11(4)18,
12(1)2,
12(1)4,
12(2)5,
12(2)7,
12(3)9,
12(3)10,
12(3)11,
12(4)15,
12(4)16,
13(1)3,
13(2)6,
13(2)7,
13(2)9,
13(3)14
- size,
5(3)245,
6(3)9,
9(3)12
- speech,
1(1)83,
4(1)38,
6(3)9,
7(1)2,
7(3)10,
8(1)2,
8(1)4,
8(4)14,
8(4)18,
9(1)2,
10(1)6,
10(2)7,
11(1)2,
11(3)10,
13(4)16
- statistical,
1(1)3,
3(2)87,
3(4)243,
5(2)121,
5(4)323,
5(4)360,
6(1)z-4,
7(1)1,
8(1)2,
8(1)4,
8(2)6,
8(2)7,
8(2)8,
8(2)9,
8(3)10,
8(4)15,
8(4)19,
9(2)6,
9(3)11,
10(4)18,
11(2)6,
11(2)7,
11(3)8,
11(4)15,
12(1)1,
12(3)12,
12(4)14,
12(4)16,
12(4)17,
13(1)2,
13(1)3,
13(1)4,
13(4)17
- study,
4(2)159,
4(3)243,
5(2)121,
5(2)146,
5(2)165,
5(3)209,
6(2)6,
6(2)7,
8(1)3,
8(1)4,
8(4)16,
9(2)5,
9(2)6,
9(3)11,
10(2)10,
10(3)12,
10(4)17,
10(4)18,
11(1)3,
11(2)6,
11(3)9,
11(3)11,
11(4)14,
13(1)3,
13(2)7,
13(3)11,
13(3)12,
13(3)14
- than,
5(2)146,
6(3)10,
7(3)8,
7(4)11,
7(4)13,
8(2)8,
8(4)16,
9(1)2,
9(3)11,
9(3)12,
10(1)3,
10(1)4,
10(3)14,
11(2)4,
11(2)7,
11(3)8,
11(3)9,
11(4)13,
11(4)15,
12(2)7,
12(3)10,
12(4)16,
13(1)1,
13(4)17
- then,
5(2)121,
6(2)6,
7(1)1,
7(3)10,
7(4)12,
8(1)4,
8(2)7,
8(3)10,
8(3)11,
8(3)12,
8(4)14,
9(1)1,
9(3)11,
10(2)7,
10(3)13,
10(4)20,
11(1)3,
11(2)7,
11(3)11,
11(4)15,
12(1)3,
12(3)10,
12(4)17,
13(1)4,
13(2)9,
13(3)13,
13(4)16
- thus,
6(2)6,
7(4)11,
10(2)9,
12(1)1,
12(4)15
- topic,
2(2)85,
3(4)227,
5(4)388,
6(3)11,
8(1)2,
8(3)12,
9(3)12,
9(4)15,
11(2)4,
11(2)5
- tried,
13(2)8
- Unlike,
8(2)6
- unsupervised,
6(2)6,
7(3)9,
8(1)3,
9(1)3,
9(1)4,
10(2)7,
11(3)9,
13(3)12
- validate,
6(2)6
- voting,
6(4)3,
10(2)9
- was,
5(2)146,
6(3)9,
6(3)11,
8(4)18,
8(4)19,
9(3)10,
10(4)18,
11(1)2,
11(3)10,
12(4)14,
13(1)4,
13(4)16
- weighted,
10(2)9
- were,
6(3)9,
10(1)2,
10(2)10,
11(3)10,
11(4)13,
12(4)17,
13(2)6,
13(4)16
- where,
7(3)8,
7(3)9,
7(3)10,
8(2)7,
8(2)8,
9(1)3,
9(4)13,
10(4)18,
11(2)4,
11(4)16,
12(1)3,
12(4)14
- window,
10(2)10
- word-based,
1(3)173,
1(4)297,
6(3)9,
9(3)11,
11(2)7
- would,
8(4)14,
9(3)11,
10(1)5,
12(3)9,
12(3)11
- year,
7(4)11,
8(4)18,
11(4)15