應用分群概念之改良式混合推薦機制

A Modified Hybrid Recommendation Mechanism using clustering concept

鄭伊珊(Yi-Shan Cheng)

指導教授 ： 林志浩

中原大學/商學院/資訊管理研究所/碩士(2010年)

https://doi.org/10.6840/cycu201000918

摘要

在資訊爆炸的時代裡，過多的資訊圍繞著人們周圍。如何協助使用者做資訊篩選，並提高人們獲得所需資訊的效率，是非常重要的。因此，研究中加入一些使用者個人資訊，並利用分類與模擬的工具來建構使用者偏好模型，預測他們可能感興趣的項目。研究目的是希望盡可能的貼近使用者偏好，推薦使用者真正需要的內容與項目。
首先，本研究運用k-means集群演算法將屬性相關的使用者進行分群，再使用類神經網路訓練每個群體的網路並模擬使用者評比。另一方面，模糊推薦方法藉由尋找與使用者志趣相投的鄰居，參考鄰居的喜好進行推薦。最後，結合k-means集群、類神經網路、以及模糊推薦的方法，對使用者進行推薦。
為了改善傳統推薦方法所遭受的新使用者問題，本研究提出的方法在新使用者沒有評比可輸入網路中訓練時，藉由同一群體裡其他使用者的評比建構網路後，再模擬新使用者的預測評比，藉此可以改善新使用者問題。本研究所提出的方法相較於類神經網路推薦、決策樹、關連規則等演算法，實驗後可得知有較佳的預測準確性，進而增加推薦結果的品質。

關鍵字

模糊集合 ； 類神經網路 ； k-平均法 ； 個人化推薦 ； 混合過濾

並列摘要

In the era of information explosion, a lot of information surrounds our daily life. It is important that helping people to filter out unnecessary data can improve their performance on obtaining appropriate information. Therefore, this study adopts some user profile information to construct user preference model. This research also develops a classified method and a simulated tool to recommend items and contents for users.
Firstly, the proposed method uses k-means clustering method to group users according to their personal attributes. Secondly, we use neural networks to simulate user’s preference. On the other hand, fuzzy method considers the preferences of users to recommend items by searching through neighborhood. Finally, this system combines k-means clustering, neural networks, and fuzzy methods to recommended items for users.
To resolve the new user problem of traditional recommendation methods, the proposed method uses the rating results of existing neighbors in the same cluster to construct the preference network of new users to predict user’s rating results. Comparing the experimental results obtained from neural networks, decision tree, and association rules, the proposed method can achieve better prediction accuracy and increase the quality of recommendation results.

並列關鍵字

K-means ； Neural network ； Hybrid filtering ； Personalized recommendation ； Fuzzy sets

參考文獻

[1]M. Y. H. Al-Shamri and K. K. Bharadwaj, "Fuzzy-genetic approach to recommender systems based on a novel hybrid user model," Expert Systems with Applications, vol. 35, pp. 1386-1399, 2008.

[2]Y. Li, L. Lu, and L. Xuefeng, "A hybrid collaborative filtering method for multiple-interests and multiple-content recommendation in E-Commerce," Expert Systems with Applications, vol. 28, no. 1, pp. 67-77, 2005.

[3]G. Adomavicius and A. Tuzhilin, "Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions," IEEE Transactions on Knowledge and Data Engineering, vol. 17, pp. 734-749, June 2005.

[4]Y. Dai, H. Ye, and S. Gong, "Personalized Recommendation Algorithm Using User Demography Information," In Proc. International Workshop on Knowledge Discovery and Data Mining, pp. 100-103, 2009.

[6]D. Goldberg, D. Nichols, B. M. Oki, and D. Terry, "Using collaborative filtering to weave an information tapestry," Communications ACM, vol. 35, no. 12, pp. 61-70, 1992.