在Web2.0 的熱潮下,個人從資訊的閱讀者變成創造者,資訊量快速地成長,社群媒體(social media)的推薦充滿挑戰。隨著標註(tagging)這個機制被應用在許多社群媒體網站上,本研究提出一個新穎的方法,從使用者個人收藏中所擁有的標籤(tags),來了解使用者的興趣,產生使用者描述(user profile)。而在標註機制中所擁有的一個重要特性,社群互動(social interaction),讓我們可以進一步地拓展成從朋友的角度來了解使用者興趣。 有別於傳統的推薦系統,在缺乏使用者明確評分(ratings)的社群媒體上,我們定義一個標籤對標籤(tag-to-tag)的矩陣,來達到協同式過濾(collaborative filtering) 推薦的效果。並且考慮使用者興趣間的關聯程度,做出具有主題關聯性的推薦。我們收集del.icio.us 書籤網站上17,435 個使用者的資料,評估產生的使用者描述與推薦的效果。另外,更實際設計了小規模的使用者調查,了解他們對於推薦的滿意程度,調查結果顯示大約百分之六十的推薦能被使用者所接受。
Making recommendations for social media presents special challenges. As the fact of tagging becomes a common practice at many social media sites, this research proposes a new approach to user profiling based on the tags associated with one’s personal collection of contents. To utilize the social interaction inherent in tagging, a personal profile can be further extended with the tags specified by one’s social contacts. A tag-to-tag matrix is defined to enable collaborative filtering-styled recommendations without explicit user ratings. By looking up this matrix and considering the user’s conceptual associations, we presented the process of making social media recommendations based on the user’s tag-based profile or based on a given tag. A practical implementation is applied on the del.icio.us bookmarks and tags of 17,435 users, and both user profiles and recommendations are generated and evaluated. Our small-scaled user study shows that most of our recommendations satisfied the testers.