為增進知識被分享的能力,線上知識論壇被廣泛的使用在知識管理實務以及線上社群中。此外對於某個特定的主題下,能夠認定專家的能力是增進知識管理能力的方法之一。現存的專家認定技術可以大約分為兩大類,第一類是以內容為基礎的專家認定系統,另外一類則為以訊息交流為基礎的專家認定系統。然而,對於線上論壇的專家認定議題,這兩種系統有存在著許多的局限。由此,本研究提出了一個基於論壇中參予者評價來認定專家的方法,更明確的說,我們延伸了PageRank(一種基於圖型結構的排名調整機制,常被用於以訊息交流為基礎的專家認定系統)提出了一個ExpertRank的演算法,能夠同時顧及網路論壇成員互動的正面及負面意見對於專家認定的影響。我們從一個產品評價論壇網站(Epinions.com)取得了三個資料集合,結果發現我們提出的ExpertRank能夠得到比PageRank更好的準確度。
Online forums have been extensively used in many knowledge management practices as well as online communities for sharing knowledge. Identifying who are experts of certain topics is essential to effective knowledge sharing in online forums. Existing expert identification techniques can broadly be classified into two major categories: content-based and communication-based expert identification techniques. However, they incur several limitations when applying to expert identification from online forums. In this study, we propose an expert identification technique on the basis of opinion ratings by members in online forums. Specifically, we extend PageRank, a graph-based ranking mechanism commonly employed by existing communication-based expert identification techniques and propose an ExpertRank algorithm that considers both positive and negative opinion ratings in communications in online forums. Using three datasets collected from a product review website (ie.., Epinions.com), our empirical evaluation results show that our proposed ExpertRank algorithm outperforms PageRank.