The traditional recommender systems are usually oriented to general situations in daily lives (e.g. recommend movies, books, music, news and etc.), but seldom cover the recommendation scenarios for the collaborative team environments. We have done an explorative study on collaborative filtering mechanism for collaborative team environments, which is some kind of multi-dimensional recommender systems problem with consideration of workflow context. This paper proposed 3-dimensional workflow space model, and investigated the new similarities measure between members in workflow space. Then, the new similarities measure is utilized into collaborative filtering for recommender systems in collaborative team environments. At last, the efficiency and usability of the proposed method are validated by experiments referring to a real-world collaborative team of a manufacturing enterprise.
With the advancement of electronic commerce, recommender systems (RSs) have become an important research area, since the appearance of the first paper on collaborative filtering in mid-1990s (Hill et al., 1995, Resnick et al., 1994 and Shardanand and Maes, 1995). There have been numerous ways of recommendation methods that utilize various types of data and analysis tools (Adomavicius and Tuzhilin, 2005 and Burke, 2002). One of the most famous methods is collaborative filtering (CF), which recommend the user with the items that people with similar tastes and preferences liked in the past (Herlocker, Konstan, Terveen, & Riedl, 2004). The collaborative filtering has been validated to be successful and efficient by a large number of researches in academia area, and also has been implemented by many real-world businesses in industrial area.
Those traditional CF-based recommender systems are usually oriented to general situations in daily lives (e.g. recommend movies, books, music, news and etc.). Few existing researches have been covered specific situations, such as collaborative environments. Team members among a collaborative team usually come from diverse disciplines, each with particular expertise and contribution from their relevant areas. So their demands for knowledge are also different from each other (Zhen & Jiang, 2008). Collaborative team also requires a mechanism to efficiently recommend proper knowledge to proper members.
The CF mechanism in collaborative team environments will be different from traditional CF-based recommender systems in daily lives. The CF mechanism is essentially heuristics that make rating predictions based on similar members, so the most important step for the collaborative filtering mechanism is finding similarities between users effectively. The traditional CF so far has relied on vector similarity measures (e.g. Pearson’s correlation or Cosine) of existing user-item rating records. As to the collaborative team environments with a specific domain, the traditional CF is not most suitable and efficient, especially when the existing user-item rating records are not sufficient. Moreover, the inner relationships involved in collaborative team environments are ignored by traditional CF in calculating the similarities between users, which refers to the serious degradation of recommendation quality.
This paper has been done an explorative study on CF mechanism in collaborative team environments by taking some domain-specific context information into account for CF. More specifically; we view CF as an organizational process and investigate process-oriented solution for it. As the premise of CF, the recommendation entities are also set with the orientation for organizational process, such as workflow. Workflow model consists of three key concepts: members, roles and tasks. Therefore, the CF problem in this paper is involved in a multi-dimensional space: (members × roles × tasks × items) rather than ordinary 2-dimension problem: (members × items). A workflow space model is proposed in this paper, and is utilized to solve the multi-dimensional CF problem for collaborative team environment. This paper mainly studied the similarities measure between members based on the proposed workflow space model, and also utilized the similarities measure into CF in experiments, which refer to a real-world collaborative team of a manufacturing company.
The rest of this paper is organized as follows. Some related works done by other scholars are briefly introduced in the next section. In Section 3, we introduce multi-dimensional recommender systems within workflow context. The workflow space model is proposed and introduced in Section 4. As to the applications of workflow space in collaborative filtering, Section 5 gives detail illustrations. Section 6 is the experimental evaluation for the proposed approaches. Closing remarks and summary are then outlined in the last section.
