Increasing Topic Coherence by Aggregating Topic Models

Stuart Blair, Yaxin Bi, Maurice Mulvenna

Research output: Chapter in Book/Report/Conference proceedingChapter

1 Citation (Scopus)

Abstract

In this paper, we introduce a novel method for aggregating multiple topic models to produce an aggregate model that contains top- ics with greater coherence than individual models. When generating a topic model a number of parameters must be specified. Depending on the parameters chosen the resulting topics can be very general or very specific. In this paper the process of aggregating multiple topic mod- els generated using different parameters is investigated; the hypothesis being that combining the general and specific topics can increase topic coherence. The aggregate model is created using cosine similarity and Jensen-Shannon divergence to combine topics which are above a sim- ilarity threshold. The model is evaluated using evaluation methods to calculate the coherence of topics in the base models against those of the aggregated model. The results presented in this paper show that the aggregated model outperforms standard topic models at a statistically significant level in terms of topic coherence when evaluated against an external corpus.
LanguageEnglish
Title of host publicationProceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016)
EditorsF Lehner, N Fteimi
Place of PublicationHeidelberg
Pages69-81
Volume9983
DOIs
Publication statusE-pub ahead of print - 5 Oct 2016

Keywords

  • Topic models
  • Semantic coherence
  • Ensemble methods

Cite this

Blair, S., Bi, Y., & Mulvenna, M. (2016). Increasing Topic Coherence by Aggregating Topic Models. In F. Lehner, & N. Fteimi (Eds.), Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016) (Vol. 9983, pp. 69-81). Heidelberg. https://doi.org/10.1007/978-3-319-47650-6_6
Blair, Stuart ; Bi, Yaxin ; Mulvenna, Maurice. / Increasing Topic Coherence by Aggregating Topic Models. Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016). editor / F Lehner ; N Fteimi. Vol. 9983 Heidelberg, 2016. pp. 69-81
@inbook{d62c71c558364b3aa9cd821495cbfef4,
title = "Increasing Topic Coherence by Aggregating Topic Models",
abstract = "In this paper, we introduce a novel method for aggregating multiple topic models to produce an aggregate model that contains top- ics with greater coherence than individual models. When generating a topic model a number of parameters must be specified. Depending on the parameters chosen the resulting topics can be very general or very specific. In this paper the process of aggregating multiple topic mod- els generated using different parameters is investigated; the hypothesis being that combining the general and specific topics can increase topic coherence. The aggregate model is created using cosine similarity and Jensen-Shannon divergence to combine topics which are above a sim- ilarity threshold. The model is evaluated using evaluation methods to calculate the coherence of topics in the base models against those of the aggregated model. The results presented in this paper show that the aggregated model outperforms standard topic models at a statistically significant level in terms of topic coherence when evaluated against an external corpus.",
keywords = "Topic models, Semantic coherence, Ensemble methods",
author = "Stuart Blair and Yaxin Bi and Maurice Mulvenna",
year = "2016",
month = "10",
day = "5",
doi = "10.1007/978-3-319-47650-6_6",
language = "English",
isbn = "978-3-319-47650-6",
volume = "9983",
pages = "69--81",
editor = "F Lehner and N Fteimi",
booktitle = "Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016)",

}

Blair, S, Bi, Y & Mulvenna, M 2016, Increasing Topic Coherence by Aggregating Topic Models. in F Lehner & N Fteimi (eds), Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016). vol. 9983, Heidelberg, pp. 69-81. https://doi.org/10.1007/978-3-319-47650-6_6

Increasing Topic Coherence by Aggregating Topic Models. / Blair, Stuart; Bi, Yaxin; Mulvenna, Maurice.

Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016). ed. / F Lehner; N Fteimi. Vol. 9983 Heidelberg, 2016. p. 69-81.

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Increasing Topic Coherence by Aggregating Topic Models

AU - Blair, Stuart

AU - Bi, Yaxin

AU - Mulvenna, Maurice

PY - 2016/10/5

Y1 - 2016/10/5

N2 - In this paper, we introduce a novel method for aggregating multiple topic models to produce an aggregate model that contains top- ics with greater coherence than individual models. When generating a topic model a number of parameters must be specified. Depending on the parameters chosen the resulting topics can be very general or very specific. In this paper the process of aggregating multiple topic mod- els generated using different parameters is investigated; the hypothesis being that combining the general and specific topics can increase topic coherence. The aggregate model is created using cosine similarity and Jensen-Shannon divergence to combine topics which are above a sim- ilarity threshold. The model is evaluated using evaluation methods to calculate the coherence of topics in the base models against those of the aggregated model. The results presented in this paper show that the aggregated model outperforms standard topic models at a statistically significant level in terms of topic coherence when evaluated against an external corpus.

AB - In this paper, we introduce a novel method for aggregating multiple topic models to produce an aggregate model that contains top- ics with greater coherence than individual models. When generating a topic model a number of parameters must be specified. Depending on the parameters chosen the resulting topics can be very general or very specific. In this paper the process of aggregating multiple topic mod- els generated using different parameters is investigated; the hypothesis being that combining the general and specific topics can increase topic coherence. The aggregate model is created using cosine similarity and Jensen-Shannon divergence to combine topics which are above a sim- ilarity threshold. The model is evaluated using evaluation methods to calculate the coherence of topics in the base models against those of the aggregated model. The results presented in this paper show that the aggregated model outperforms standard topic models at a statistically significant level in terms of topic coherence when evaluated against an external corpus.

KW - Topic models

KW - Semantic coherence

KW - Ensemble methods

U2 - 10.1007/978-3-319-47650-6_6

DO - 10.1007/978-3-319-47650-6_6

M3 - Chapter

SN - 978-3-319-47650-6

VL - 9983

SP - 69

EP - 81

BT - Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016)

A2 - Lehner, F

A2 - Fteimi, N

CY - Heidelberg

ER -

Blair S, Bi Y, Mulvenna M. Increasing Topic Coherence by Aggregating Topic Models. In Lehner F, Fteimi N, editors, Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management (KSEM-2016). Vol. 9983. Heidelberg. 2016. p. 69-81 https://doi.org/10.1007/978-3-319-47650-6_6