Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions

Gabriella Skitalinskaya; Mikhail Alexandrov; John Cardiff

doi:10.1007/978-3-030-02840-4_3

Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions

Gabriella Skitalinskaya, Mikhail Alexandrov, John Cardiff

School of Enterprise Computing and Digital Transformation

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In this paper we present a two-pass algorithm based on different matrix decompositions, such as LSI, PCA, ICA and NMF, which allows tracking of the evolution of topics over time. The proposed dynamic topic models as output give an easily interpreted overview of topics found in a sequentially organized set of documents that does not require further processing. Each topic is presented by a user-specified number of top-terms. Such an approach to topic modeling if applied to, for example, a news article data set, can be convenient and useful for economists, sociologists, political scientists. The proposed approach allows to achieve results comparable to those obtained using complex probabilistic models, such as LDA.

Original language	English
Title of host publication	Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings
Editors	Miguel González-Mendoza, Félix Castro, Sabino Miranda-Jiménez
Publisher	Springer Verlag
Pages	27-43
Number of pages	17
ISBN (Print)	9783030028398
DOIs	https://doi.org/10.1007/978-3-030-02840-4_3
Publication status	Published - 2018
Event	16th Mexican International Conference on Artificial Intelligence, MICAI 2017 - Enseneda, Mexico Duration: 23 Oct 2017 → 28 Oct 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10633 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th Mexican International Conference on Artificial Intelligence, MICAI 2017
Country/Territory	Mexico
City	Enseneda
Period	23/10/17 → 28/10/17

Keywords

Dynamic topic modeling
Latent Dirichlet allocation
Matrix decomposition

Access to Document

10.1007/978-3-030-02840-4_3Licence: CC BY

https://arrow.tudublin.ie/ittscicon/11Licence: CC BY

Cite this

Skitalinskaya, G., Alexandrov, M., & Cardiff, J. (2018). Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions. In M. González-Mendoza, F. Castro, & S. Miranda-Jiménez (Eds.), Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings (pp. 27-43). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10633 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-02840-4_3

Skitalinskaya, Gabriella ; Alexandrov, Mikhail ; Cardiff, John. / Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions. Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings. editor / Miguel González-Mendoza ; Félix Castro ; Sabino Miranda-Jiménez. Springer Verlag, 2018. pp. 27-43 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{55a5d08a63174ed78745cd79d6da2564,

title = "Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions",

abstract = "In this paper we present a two-pass algorithm based on different matrix decompositions, such as LSI, PCA, ICA and NMF, which allows tracking of the evolution of topics over time. The proposed dynamic topic models as output give an easily interpreted overview of topics found in a sequentially organized set of documents that does not require further processing. Each topic is presented by a user-specified number of top-terms. Such an approach to topic modeling if applied to, for example, a news article data set, can be convenient and useful for economists, sociologists, political scientists. The proposed approach allows to achieve results comparable to those obtained using complex probabilistic models, such as LDA.",

keywords = "Dynamic topic modeling, Latent Dirichlet allocation, Matrix decomposition",

author = "Gabriella Skitalinskaya and Mikhail Alexandrov and John Cardiff",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2018.; 16th Mexican International Conference on Artificial Intelligence, MICAI 2017 ; Conference date: 23-10-2017 Through 28-10-2017",

year = "2018",

doi = "10.1007/978-3-030-02840-4\_3",

language = "English",

isbn = "9783030028398",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "27--43",

editor = "Miguel Gonz{\'a}lez-Mendoza and F{\'e}lix Castro and Sabino Miranda-Jim{\'e}nez",

booktitle = "Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings",

address = "Germany",

}

Skitalinskaya, G, Alexandrov, M & Cardiff, J 2018, Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions. in M González-Mendoza, F Castro & S Miranda-Jiménez (eds), Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10633 LNAI, Springer Verlag, pp. 27-43, 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Enseneda, Mexico, 23/10/17. https://doi.org/10.1007/978-3-030-02840-4_3

Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions. / Skitalinskaya, Gabriella; Alexandrov, Mikhail; Cardiff, John.
Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings. ed. / Miguel González-Mendoza; Félix Castro; Sabino Miranda-Jiménez. Springer Verlag, 2018. p. 27-43 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10633 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions

AU - Skitalinskaya, Gabriella

AU - Alexandrov, Mikhail

AU - Cardiff, John

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2018.

PY - 2018

Y1 - 2018

N2 - In this paper we present a two-pass algorithm based on different matrix decompositions, such as LSI, PCA, ICA and NMF, which allows tracking of the evolution of topics over time. The proposed dynamic topic models as output give an easily interpreted overview of topics found in a sequentially organized set of documents that does not require further processing. Each topic is presented by a user-specified number of top-terms. Such an approach to topic modeling if applied to, for example, a news article data set, can be convenient and useful for economists, sociologists, political scientists. The proposed approach allows to achieve results comparable to those obtained using complex probabilistic models, such as LDA.

AB - In this paper we present a two-pass algorithm based on different matrix decompositions, such as LSI, PCA, ICA and NMF, which allows tracking of the evolution of topics over time. The proposed dynamic topic models as output give an easily interpreted overview of topics found in a sequentially organized set of documents that does not require further processing. Each topic is presented by a user-specified number of top-terms. Such an approach to topic modeling if applied to, for example, a news article data set, can be convenient and useful for economists, sociologists, political scientists. The proposed approach allows to achieve results comparable to those obtained using complex probabilistic models, such as LDA.

KW - Dynamic topic modeling

KW - Latent Dirichlet allocation

KW - Matrix decomposition

UR - https://www.scopus.com/pages/publications/85059954699

U2 - 10.1007/978-3-030-02840-4_3

DO - 10.1007/978-3-030-02840-4_3

M3 - Conference contribution

AN - SCOPUS:85059954699

SN - 9783030028398

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 27

EP - 43

BT - Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings

A2 - González-Mendoza, Miguel

A2 - Castro, Félix

A2 - Miranda-Jiménez, Sabino

PB - Springer Verlag

T2 - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017

Y2 - 23 October 2017 through 28 October 2017

ER -

Skitalinskaya G, Alexandrov M, Cardiff J. Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions. In González-Mendoza M, Castro F, Miranda-Jiménez S, editors, Advances in Computational Intelligence - 16th Mexican International Conference on Artificial Intelligence, MICAI 2017, Proceedings. Springer Verlag. 2018. p. 27-43. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-02840-4_3

Comparison of two-pass algorithms for dynamic topic modeling based on matrix decompositions

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this