推薦策略與推薦資訊對線上推薦績效影響之研究

A Study for the Effect of the Recommendation Strategy and Information on the on-line Recommendation Performance

吳晨帆(Chen-Fan Wu)

指導教授 ： 吳肇銘

中原大學/商學院/資訊管理研究所/碩士(2005年)

https://doi.org/10.6840/cycu200500564

摘要

觀察實務環境，雖然許多電子商務網站已成功利用推薦系統及其機制創造商機與競爭優勢，但是推薦系統的研究似乎都沒有充分強調並反映出商務部分的特色與議題，Haubl & Murray(2005)認為，推薦系統(推薦代理人)基本上扮演著雙重角色，一面協助顧客，一方面則是影響改變顧客的決策。既然推薦系統被認為是解決電子商務環境下資訊超載、有效協助顧客制定購買決策的利器的話，應從消費者決策面切入，以獲得提升推薦系統績效與能力之線索。

本研究試圖以消費者決策的角度重新思考推薦系統績效之議題，特別是運用在電子商務環境中的推薦系統，探討並了解「推薦策略」與「推薦資訊」對於「推薦績效」的不同影響？而在網站經營業者有其考量下，必須採用某些「特定的推薦策略」來服務顧客時，是否能利用「特定的推薦資訊」配合來提升推薦的績效？根據實地實驗資料分析，歸納出以下的研究結論：
一、 「顧客導向」與「市場導向」推薦策略，對消費者的需求具有高「命中度」，同時推薦的手機使消費者具有高「購買意願」。由此可證推薦系統的演算法以及使用的資料集，將對推薦績效有絕對影響力。

二、 不同「推薦策略」或是「推薦資訊」對「滿意度」的影響並不顯著。適當地與適量地給予消費者所需的資訊，使其更容易地作出購買決策，避免自行收集並評估大量資訊而產生「資訊超載」的狀況，對其資訊蒐集上產生一定程度的幫助，自然而然消費者對的滿意度就會正向提高許多。

三、 「專家測試報告」有助於「命中度」的提升；而「討論區文章」效果最差。網路商店經營者在其網路購物環境的建置中，提供推薦資訊是且必要的但是在資訊的類型則必須有選擇考量。

本研究除了提供上述建議给網站經營業者參考，在研究貢獻部份除了驗證不同推薦策略與推薦資訊對推薦績效的影響之外，同時也建議了適用於資訊提供型網站的推薦機制，此外也補強了推薦系統在消費者決策相關研究的領域，提供後續推薦系統研究者一個不同的研究方向。

關鍵字

推薦資訊 ； 推薦系統 ； 電子商務

並列摘要

Observed the actual practice circumstance, most of the large e-commerce websites, such as Amazon.com, have successfully adopted the recommender system and its mechanism to initiate business opportunities and competitive edges. However, it seems that until now past researches about the recommender system do not fully emphasize and reflect issues in commerce. Haubl & Murray (2005) pointed out that the recommender system (or recommender agents) basically plays a dual role. It not only assists the consumers but also affect their purchasing decisions. Since recommender systems are regarded as the best tool to solve information overloading and help consumers to make effective purchasing decisions. It is possible to discover new ideas of performance promotion.
From perspective of consumer purchasing decision stages, our research tries to aim at the issue of the performance of the recommender system; especially those utilized in e-commerce environment. We try to understand influence of different recommendation strategies and information on recommendation performance. Also we try to find out if there existed any manners to utilize specific recommendation information to perform better result when proprietors have there own specific recommendation strategies. Based on statistics of field experiment, the results are as followings:
1. The “costumer-oriented” and “market-oriented” recommendation strategies perform high “precision” and result in higher purchasing intension. This shows the algorithm of the recommender system and the dataset have absolute influence on the recommendation performance.
2. There is no significant difference on the influence of different “recommender strategy” or “recommender information” to satisfaction. Appropriately and suitably provide recommender information for the consumers will help them to make purchasing decisions more easily. By this way, consumers could avoid facing information overloading led from information collected and evaluation and spontaneously prompt the satisfaction to the positive quadrant.
3. “Experts testing reports” helps to raise the satisfaction, but on the contrary, the “articles adopted from the forum” result worst. It is absolute necessary and essential to provide recommendation information in the online shopping environment. However proprietors have to be cautious about the information type they offering.
The research gives suggestions to website proprietors as reference. In the research contribution, the results of the influence of different “recommender strategy” or “recommender information” to effects are tested and verified as well as suggests the recommender mechanism for the informative website. Besides, the research also renews the domain of the recommender system and consumers’ decision, and supplies follow-up researchers a distinct object of study.

並列關鍵字

Recommender system ； Electronic commerce ； Recommender information

參考文獻

5. 高振智，「以消費者購買決策為基礎之適性化推薦系統」，中原大學資訊管理研究所，碩士論文，民國92年。

6. 陳素汝，「網際網路上搜尋行為之分析與實驗」，中原大學國際貿易研究所，碩士論文，民國九十二年七月。

13. 劉韋緒，「資訊代理人應用於行動化顧客關係管理之研究」，中原大學資訊管理研究所，碩士論文，民國90年。

12. 廖婉菁，「應用協同過濾機制於商品推薦之研究 —以手機網站為例」，中原大學資訊管理研究所，碩士論文，民國90年。

4. B. Sarwar, J. Konstan, A. Borchers, J. Herlocker, B. Miller and J. Riedl, ”Using Filtering Agents to Improve Prediction Quality in the GroupLens Research Collaborative Filtering System”, Proceedings of the 1998 Conference on Computer Supported Cooperative Work, 1998.