融合多模型排序之點擊預測模型

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：27

、訪客IP：3.147.140.101

姓名

賴品菁(Pin-Ching Lai) 查詢紙本館藏

畢業系所

資訊工程學系

論文名稱

融合多模型排序之點擊預測模型
(Learning to ensemble ranking model for sequential click prediction)

相關論文

★ 透過網頁瀏覽紀錄預測使用者之個人資訊與性格特質	★ 透過矩陣分解之多目標預測方法預測使用者於特殊節日前之瀏覽行為變化
★ 動態多模型融合分析研究	★ 擴展點擊流：分析點擊流中缺少的使用者行為
★ 關聯式學習：利用自動編碼器與目標傳遞法分解端到端倒傳遞演算法	★ 分析網路日誌中有意圖、無意圖及缺失之使用者行為
★ 基於自注意力機制產生的無方向性序列編碼器使用同義詞與反義詞資訊調整詞向量	★ 探索深度學習或簡易學習模型在點擊率預測任務中的使用時機
★ 空氣品質感測器之故障偵測--基於深度時空圖模型的異常偵測框架	★ 以同反義詞典調整的詞向量對下游自然語言任務影響之實證研究
★ 結合時空資料的半監督模型並應用於PM2.5空污感測器的異常偵測	★ 藉由權重之梯度大小調整DropConnect的捨棄機率來訓練神經網路
★ 使用圖神經網路偵測 PTT 的低活躍異常帳號	★ 針對個別使用者從其少量趨勢線樣本生成個人化趨勢線
★ 基於雙變量及多變量貝他分布的兩個新型機率分群模型	★ 一種可同時更新神經網路各層網路參數的新技術— 採用關聯式學習及管路化機制

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

現今因為網路的蓬勃發展，網路與人們的生活密不可分，在網站的
需求上，不管是購物網站或是音樂、影片網站，越來越多網站有個人化
的服務及推薦。若能有效的推薦商品給使用者，不但能增加使用者對此
網站的滿意度，還能讓這個網站的創造者獲得收益，造就雙贏的局面。
本篇論文利用兩種不同大小的電商資料，對使用者下一個可能會點
擊的商品做排序，推薦給使用者。為了將真正被點擊過的商品排在更前
面的推薦序列中，我們使用了四種不同類型的模型來做排序。由於最近
深度學習在各個領域表現卓越，推薦系統也開始使用深度學習架構來訓
練，但我們發現，在不同的電商規模下，使用深度學習架構不一定會表
現得比淺層的模型好。
在發現了這個現象之後，我們使用這四個不同的模型排序過後的序
列排名，來當作我們的訓練資料，並重新排序，最後我們得到了更符合使
用者歷史點擊的排名，將真正被點擊過的商品的排名往前排，並獲得明
顯的進步。不管在哪個資料集上，都表現的比四個基準模型還要好，並
在實驗過程中發現了商品排名與商品出現次數的相關性。

摘要(英)

With the vigorous development of the Internet, it becomes inseparable
from people’s lives. Various websites, such as online retailers and online
music/video streaming websites, provide personalized recommendations to
show the RIGHT products to users. These recommendations may increase
users’ satisfaction with these services and create revenues for the service
providers.
This paper aims to predict a users’ next clicking item based on two
e-commerce datasets. We compared four ranking models and claimed a
model performs better if the model ranks the next clicking item at a former
position among the four models. Despite recent researches showed excellent
performance on deep learning-based ranking models, we found that this is
not always the case. Notably, we may need to consider the size of the
service providers as a reference to decide to apply a deep or a shallow
learning model.
Since different models may work under various scenarios, we developed
an ensemble model that learns to rank the items based on the rankings
returned by the four ranking models. Experimental results show that the
new model outperforms the four baselines: on both datasets, the new model
tends to put the next clicking item at a former position.

關鍵字(中)

★ Top-N排序
★ 推薦系統
★ 點擊率預測
★ 機器學習
★ 深度學習

關鍵字(英)

論文目次

摘要ix
Abstract xi
目錄xiii
一、緒論1
1.1 研究動機.................................................................. 1
1.2 研究目標.................................................................. 2
1.3 研究貢獻.................................................................. 2
1.4 論文架構.................................................................. 3
二、背景及相關論文5
2.1 推薦系統Top-N 名候選商品排名.................................. 5
2.2 簡易模型與深度學習於推薦系統之應用........................... 7
2.3 深度學習模型是否確實提升推薦系統效果........................ 9
三、實驗發想與方法11
3.1 候選商品召回與商品向量訓練....................................... 11
3.2 多層神經網路(Multilayer perceptron) ........................ 12
3.3 阿里巴巴集團提出之深度學習模型................................. 13
3.3.1 Deep Interest Network (DIN) ......................... 14
3.3.2 Deep Interest Evolution Network (DIEN)........ 16
xiii
目錄
3.4 融合多模型排序結果................................................... 17
四、實驗設置與結果19
4.1 資料集介紹............................................................... 19
4.1.1 電商A 資料集................................................. 20
4.1.2 阿里巴巴集團旗下淘寶資料集.............................. 20
4.2 實驗資料配置與前處理................................................ 21
4.2.1 資料集介紹與前處理.......................................... 21
4.2.2 四個基礎模型實驗介紹....................................... 22
4.3 評量指標.................................................................. 24
4.4 實驗結果.................................................................. 25
4.4.1 基礎模型在對應資料集之排序結果........................ 26
4.4.2 參考基礎模型排序資訊之重新排序候選商品結果...... 30
五、結論與未來展望35
5.1 結論........................................................................ 35
5.2 未來展望.................................................................. 36
參考文獻37

參考文獻

[1] M. F. Dacrema, P. Cremonesi, and D. Jannach, “Are we really making much
progress? A worrying analysis of recent neural recommendation approaches,” CoRR,
vol. abs/1907.06902, 2019. arXiv: 1907.06902. [Online]. Available: http://arxiv.
org/abs/1907.06902.
[2] A. M. Elkahky, Y. Song, and X. He, “A multi-view deep learning approach for
cross domain user modeling in recommendation systems,” in Proceedings of the 24th
International Conference on World Wide Web, ser. WWW ’15, Florence, Italy:
International World Wide Web Conferences Steering Committee, 2015, pp. 278–
288, isbn: 9781450334693. doi: 10.1145/2736277.2741667. [Online]. Available:
https://doi.org/10.1145/2736277.2741667.
[3] D. Kim, C. Park, J. Oh, S. Lee, and H. Yu, “Convolutional matrix factorization for
document context-aware recommendation,” in Proceedings of the 10th ACM Conference
on Recommender Systems, ser. RecSys ’16, Boston, Massachusetts, USA:
Association for Computing Machinery, 2016, pp. 233–240, isbn: 9781450340359.
doi: 10.1145/2959100.2959165. [Online]. Available: https://doi.org/10.1145/
2959100.2959165.
[4] J. Manotumruksa, C. Macdonald, and I. Ounis, “A contextual attention recurrent
architecture for context-aware venue recommendation,” in The 41st International
ACM SIGIR Conference on Research Development in Information Retrieval,
ser. SIGIR ’18, Ann Arbor, MI, USA: Association for Computing Machinery, 2018,
pp. 555–564, isbn: 9781450356572. doi: 10 . 1145 / 3209978 . 3210042. [Online].
Available: https://doi.org/10.1145/3209978.3210042.
[5] N. Sachdeva, K. Gupta, and V. Pudi, “Attentive neural architecture incorporating
song features for music recommendation,” CoRR, vol. abs/1811.08203, 2018. arXiv:
1811.08203. [Online]. Available: http://arxiv.org/abs/1811.08203.
[6] Z. Sun, J. Yang, J. Zhang, A. Bozzon, L.-K. Huang, and C. Xu, “Recurrent knowledge
graph embedding for effective recommendation,” in Proceedings of the 12th
ACM Conference on Recommender Systems, ser. RecSys ’18, Vancouver, British
Columbia, Canada: Association for Computing Machinery, 2018, pp. 297–305, isbn:
9781450359016. doi: 10.1145/3240323.3240361. [Online]. Available: https://
doi.org/10.1145/3240323.3240361.
37
[7] Y. Tay, A. T. Luu, and S. C. Hui, “Translational recommender networks,” CoRR,
vol. abs/1707.05176, 2017. arXiv: 1707.05176. [Online]. Available: http://arxiv.
org/abs/1707.05176.
[8] Y. Tay, L. A. Tuan, and S. C. Hui, “Multi-pointer co-attention networks for recommendation,”
CoRR, vol. abs/1801.09251, 2018. arXiv: 1801 . 09251. [Online].
Available: http://arxiv.org/abs/1801.09251.
[9] T. X. Tuan and T. M. Phuong, “3d convolutional networks for session-based recommendation
with content features,” in Proceedings of the Eleventh ACM Conference
on Recommender Systems, ser. RecSys ’17, Como, Italy: Association for Computing
Machinery, 2017, pp. 138–146, isbn: 9781450346528. doi: 10.1145/3109859.
3109900. [Online]. Available: https://doi.org/10.1145/3109859.3109900.
[10] F. Vasile, E. Smirnova, and A. Conneau, “Meta-prod2vec - product embeddings using
side-information for recommendation,” CoRR, vol. abs/1607.07326, 2016. arXiv:
1607.07326. [Online]. Available: http://arxiv.org/abs/1607.07326.
[11] P. Covington, J. Adams, and E. Sargin, “Deep neural networks for youtube recommendations,”
in Proceedings of the 10th ACM Conference on Recommender
Systems, New York, NY, USA, 2016.
[12] S. Kabbur and G. Karypis, “Nlmf: Nonlinear matrix factorization methods for top-n
recommender systems,” vol. 2015, Dec. 2014. doi: 10.1109/ICDMW.2014.108.
[13] M. Deshpande and G. Karypis, “Item-based top-n recommendation algorithms,”
ACM Trans. Inf. Syst., vol. 22, no. 1, pp. 143–177, Jan. 2004, issn: 1046-8188.
doi: 10.1145/963770.963776. [Online]. Available: https://doi.org/10.1145/
963770.963776.
[14] B. Sarwar, G. Karypis, J. Konstan, and J. Riedl, “Item-based collaborative filtering
recommendation algorithms,” in Proceedings of the 10th International Conference
on World Wide Web, ser. WWW ’01, Hong Kong, Hong Kong: Association for Computing
Machinery, 2001, pp. 285–295, isbn: 1581133480. doi: 10.1145/371920.
372071. [Online]. Available: https://doi.org/10.1145/371920.372071.
[15] A. Tewari, J. Singh, and A. Barman, “Generating top-n items recommendation set
using collaborative, content based filtering and rating variance,” Procedia Computer
Science, vol. 132, pp. 1678–1684, Jan. 2018. doi: 10.1016/j.procs.2018.05.139.
[16] C. Cooper, S. Lee, T. Radzik, and Y. Siantos, “Random walks in recommender
systems: Exact computation and simulations,” Apr. 2014, pp. 811–816. doi: 10.
1145/2567948.2579244.
[17] P. Lops, M. de Gemmis, and G. Semeraro, “Content-based recommender systems:
State of the art and trends,” in. Jan. 2011, pp. 73–105. doi: 10.1007/978-0-387-
85820-3_3.
38
[18] B. Paudel, F. Christoffel, C. Newell, and A. Bernstein, “Updatable, accurate, diverse,
and scalable recommendations for interactive applications,” ACM Trans.
Interact. Intell. Syst., vol. 7, no. 1, Dec. 2016, issn: 2160-6455. doi: 10 . 1145 /
2955101. [Online]. Available: https://doi.org/10.1145/2955101.
[19] Q. Chen, H. Zhao, W. Li, P. Huang, and W. Ou, “Behavior sequence transformer
for e-commerce recommendation in alibaba,” CoRR, vol. abs/1905.06874, 2019.
arXiv: 1905.06874. [Online]. Available: http://arxiv.org/abs/1905.06874.
[20] D. Liang, R. G. Krishnan, M. D. Hoffman, and T. Jebara, “Variational autoencoders
for collaborative filtering.,” in WWW , P.-A. Champin, F. L. Gandon, M. Lalmas,
and P. G. Ipeirotis, Eds., ACM, 2018, pp. 689–698. [Online]. Available: http :
//dblp.uni-trier.de/db/conf/www/www2018.html#LiangKHJ18.
[21] X. He, L. Liao, H. Zhang, L. Nie, X. Hu, and T. Chua, “Neural collaborative
filtering,” CoRR, vol. abs/1708.05031, 2017. arXiv: 1708.05031. [Online]. Available:
http://arxiv.org/abs/1708.05031.
[22] H. Wang, N. Wang, and D. Yeung, “Collaborative deep learning for recommender
systems,” CoRR, vol. abs/1409.2944, 2014. arXiv: 1409.2944. [Online]. Available:
http://arxiv.org/abs/1409.2944.
[23] G. Zhou, C. Song, X. Zhu, X. Ma, Y. Yan, X. Dai, H. Zhu, J. Jin, H. Li, and K.
Gai, “Deep interest network for click-through rate prediction,” 2017.
[24] G. Zhou, N. Mou, Y. Fan, Q. Pi, W. Bian, C. Zhou, X. Zhu, and K. Gai, “Deep interest
evolution network for click-through rate prediction,” ArXiv, vol. abs/1809.03672,
2019.
[25] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser,
and I. Polosukhin, “Attention is all you need,” CoRR, vol. abs/1706.03762, 2017.
arXiv: 1706.03762. [Online]. Available: http://arxiv.org/abs/1706.03762.
[26] Y. Feng, F. Lv, W. Shen, M. Wang, F. Sun, Y. Zhu, and K. Yang, “Deep session
interest network for click-through rate prediction,” CoRR, vol. abs/1905.06482,
2019. arXiv: 1905 . 06482. [Online]. Available: http : / / arxiv . org / abs / 1905 .
06482.
[27] C. Cooper, S. Lee, T. Radzik, and Y. Siantos, “Random walks in recommender
systems: Exact computation and simulations,” Apr. 2014, pp. 811–816. doi: 10.
1145/2567948.2579244.
[28] T. Ebesu, B. Shen, and Y. Fang, “Collaborative memory network for recommendation
systems,” CoRR, vol. abs/1804.10862, 2018. arXiv: 1804.10862. [Online].
Available: http://arxiv.org/abs/1804.10862.
[29] B. Hu, C. Shi, W. Zhao, and P. Yu, “Leveraging meta-path based context for topn
recommendation with a neural co-attention model,” Aug. 2018. doi: 10.1145/
3219819.3219965.
39
[30] X. Li and J. She, “Collaborative variational autoencoder for recommender systems,”
in Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge
Discovery and Data Mining, ser. KDD ’17, Halifax, NS, Canada: Association for
Computing Machinery, 2017, pp. 305–314, isbn: 9781450348874. doi: 10.1145/
3097983. 3098077. [Online]. Available: https: // doi. org/ 10. 1145/ 3097983.
3098077.

指導教授

陳弘軒

審核日期

2020-7-20

推文