Exploring a Topical Representation of Documents for Recommendation Systems

Israel Mendonça, Antoine Trouvé, Akira Fukuda, Kazuaki Murakami

研究成果: 著書/レポートタイプへの貢献会議での発言

抄録

In this paper, we address the performance problems inherited when we use word embedding for recommendation. Free-text documents has no structural constructing rules, and are hard to model. Hence, the problem of having an accurate model, that conveys all the important information is a nontrivial problem. We convert the document to a numeric structure using word-embedding and test two document representations: one based in the center of this numeric representation and the other one based on pre-defined set of topics. We build a free text recommendation system and study how the performance, in terms of precision and recommendation time, is affected by both representations. We then vary the number of topics used to represent documents and verify the tradeoffs inherited from having a compact representation. The more compact the recommendation, the shorter the recommendation time, however more information is lost in the compactation process. We empirically test different possibilities for the topics and find an optimal point that is 3 times faster than a baseline and almost as accurate as it.

元の言語英語
ホスト出版物のタイトル2018 9th International Conference on Awareness Science and Technology, iCAST 2018
出版者Institute of Electrical and Electronics Engineers Inc.
ページ73-78
ページ数6
ISBN(電子版)9781538658260
DOI
出版物ステータス出版済み - 10 31 2018
イベント9th International Conference on Awareness Science and Technology, iCAST 2018 - Fukuoka, 日本
継続期間: 9 19 20189 21 2018

出版物シリーズ

名前2018 9th International Conference on Awareness Science and Technology, iCAST 2018

その他

その他9th International Conference on Awareness Science and Technology, iCAST 2018
日本
Fukuoka
期間9/19/189/21/18

Fingerprint

Recommender systems
performance
Recommendation system
time

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Human-Computer Interaction
  • Information Systems and Management
  • Experimental and Cognitive Psychology
  • Social Psychology
  • Communication

これを引用

Mendonça, I., Trouvé, A., Fukuda, A., & Murakami, K. (2018). Exploring a Topical Representation of Documents for Recommendation Systems. : 2018 9th International Conference on Awareness Science and Technology, iCAST 2018 (pp. 73-78). [8517192] (2018 9th International Conference on Awareness Science and Technology, iCAST 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICAwST.2018.8517192

Exploring a Topical Representation of Documents for Recommendation Systems. / Mendonça, Israel; Trouvé, Antoine; Fukuda, Akira; Murakami, Kazuaki.

2018 9th International Conference on Awareness Science and Technology, iCAST 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 73-78 8517192 (2018 9th International Conference on Awareness Science and Technology, iCAST 2018).

研究成果: 著書/レポートタイプへの貢献会議での発言

Mendonça, I, Trouvé, A, Fukuda, A & Murakami, K 2018, Exploring a Topical Representation of Documents for Recommendation Systems. : 2018 9th International Conference on Awareness Science and Technology, iCAST 2018., 8517192, 2018 9th International Conference on Awareness Science and Technology, iCAST 2018, Institute of Electrical and Electronics Engineers Inc., pp. 73-78, 9th International Conference on Awareness Science and Technology, iCAST 2018, Fukuoka, 日本, 9/19/18. https://doi.org/10.1109/ICAwST.2018.8517192
Mendonça I, Trouvé A, Fukuda A, Murakami K. Exploring a Topical Representation of Documents for Recommendation Systems. : 2018 9th International Conference on Awareness Science and Technology, iCAST 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 73-78. 8517192. (2018 9th International Conference on Awareness Science and Technology, iCAST 2018). https://doi.org/10.1109/ICAwST.2018.8517192
Mendonça, Israel ; Trouvé, Antoine ; Fukuda, Akira ; Murakami, Kazuaki. / Exploring a Topical Representation of Documents for Recommendation Systems. 2018 9th International Conference on Awareness Science and Technology, iCAST 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 73-78 (2018 9th International Conference on Awareness Science and Technology, iCAST 2018).
@inproceedings{e0baf66d8151423797ddb06c29ebefdb,
title = "Exploring a Topical Representation of Documents for Recommendation Systems",
abstract = "In this paper, we address the performance problems inherited when we use word embedding for recommendation. Free-text documents has no structural constructing rules, and are hard to model. Hence, the problem of having an accurate model, that conveys all the important information is a nontrivial problem. We convert the document to a numeric structure using word-embedding and test two document representations: one based in the center of this numeric representation and the other one based on pre-defined set of topics. We build a free text recommendation system and study how the performance, in terms of precision and recommendation time, is affected by both representations. We then vary the number of topics used to represent documents and verify the tradeoffs inherited from having a compact representation. The more compact the recommendation, the shorter the recommendation time, however more information is lost in the compactation process. We empirically test different possibilities for the topics and find an optimal point that is 3 times faster than a baseline and almost as accurate as it.",
author = "Israel Mendon{\cc}a and Antoine Trouv{\'e} and Akira Fukuda and Kazuaki Murakami",
year = "2018",
month = "10",
day = "31",
doi = "10.1109/ICAwST.2018.8517192",
language = "English",
series = "2018 9th International Conference on Awareness Science and Technology, iCAST 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "73--78",
booktitle = "2018 9th International Conference on Awareness Science and Technology, iCAST 2018",
address = "United States",

}

TY - GEN

T1 - Exploring a Topical Representation of Documents for Recommendation Systems

AU - Mendonça, Israel

AU - Trouvé, Antoine

AU - Fukuda, Akira

AU - Murakami, Kazuaki

PY - 2018/10/31

Y1 - 2018/10/31

N2 - In this paper, we address the performance problems inherited when we use word embedding for recommendation. Free-text documents has no structural constructing rules, and are hard to model. Hence, the problem of having an accurate model, that conveys all the important information is a nontrivial problem. We convert the document to a numeric structure using word-embedding and test two document representations: one based in the center of this numeric representation and the other one based on pre-defined set of topics. We build a free text recommendation system and study how the performance, in terms of precision and recommendation time, is affected by both representations. We then vary the number of topics used to represent documents and verify the tradeoffs inherited from having a compact representation. The more compact the recommendation, the shorter the recommendation time, however more information is lost in the compactation process. We empirically test different possibilities for the topics and find an optimal point that is 3 times faster than a baseline and almost as accurate as it.

AB - In this paper, we address the performance problems inherited when we use word embedding for recommendation. Free-text documents has no structural constructing rules, and are hard to model. Hence, the problem of having an accurate model, that conveys all the important information is a nontrivial problem. We convert the document to a numeric structure using word-embedding and test two document representations: one based in the center of this numeric representation and the other one based on pre-defined set of topics. We build a free text recommendation system and study how the performance, in terms of precision and recommendation time, is affected by both representations. We then vary the number of topics used to represent documents and verify the tradeoffs inherited from having a compact representation. The more compact the recommendation, the shorter the recommendation time, however more information is lost in the compactation process. We empirically test different possibilities for the topics and find an optimal point that is 3 times faster than a baseline and almost as accurate as it.

UR - http://www.scopus.com/inward/record.url?scp=85057377941&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85057377941&partnerID=8YFLogxK

U2 - 10.1109/ICAwST.2018.8517192

DO - 10.1109/ICAwST.2018.8517192

M3 - Conference contribution

T3 - 2018 9th International Conference on Awareness Science and Technology, iCAST 2018

SP - 73

EP - 78

BT - 2018 9th International Conference on Awareness Science and Technology, iCAST 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -