Perspectives to Social Acceptability in Professional Social Matching Systems
Professional Social Matching (PSM) is an understudied area in human-computer interaction, that can manifest in computer-supported networking, partnering, and grouping of people.
Supporting collaboration and networking is an essential design goal in ICT that has led to research and design of people recommender systems and social matching applications. While the majority of those focuses on dating application scenarios (e.g., Tinder) there are only a few focuses particularly on professional matchmaking (e.g., Shapr, Grip, and Brella).
Such systems utilize similarity-maximizing analytical approaches, following two social network evolution mechanisms — homophily (preference of like-minded people) and triadic closure (connecting only with people of strong ties like friends of friends). These mechanisms are detrimental to a professional context and by limiting access to new knowledge might decrease innovativeness and creativity.
We envision new computational solutions to PSM that can provide more informed (data-driven) and unexpected suggestions of collaborator candidates. For example, a system might provide the user with recommendations of people who share an interest or professional goal but who are from different disciplines or social circles or have complementary knowledge.
The ability of a system to efficiently present recommendations is dependent on the user interface solutions and interaction techniques. Therefore, the next generation of PSM should move beyond traditional list-based approach while presenting potential collaborators and we call for social graphs that visually indicate the inferred relevance of the match.
Perceived relevance refers to the degree of how recommendations in a matching system meet expectations of the user regarding internal drivers for collaboration and contextual factors. This affects the user’s attitude towards intervention of technology to the process of social matching. This could relate for example to different dimension s of similarity, logistics, network relations.
A data-driven approach to the design of people recommender systems could, for example, result in intelligent assistance for a user with making a connection between the represented content and their own needs, interests, or background. A system might help with hints regarding inferences of how a given recommendation would be relevant to the target user thus supporting decision making.
However, such non-traditional approaches bring risks of acceptance: gaining social insights from such systems will require more than just delivering efficient matchmaking mechanisms and usable interfaces. The following provides key perspectives and directions for making such systems also socially acceptable. One might question if others find it acceptable that collaboration decisions are made based on a seemingly small-minded algorithm’s recommendation.
The internal perspective refers to the user’s perceptions of the other people’s acceptance of their behavior and choices. For example, an expected design challenge relates to the user’s willingness to hand over some of their agency to a computational system in choosing with whom to collaborate.
The interpersonal perspective relates to the dynamics and norms of interpersonal interaction and social encounters. It remains an open question how to trigger and facilitate encounters between seemingly different people in a way that does not feel awkward, privacy intrusive or untrustworthy for anyone involved in the situation.
The organizational perspective is about the acceptance of such technology within a company or other organization. For example, a company’s interests might include preventing or controlling which of the individual workers can be matched with other people inside or outside the organization and for what purposes. Also, the information that is available about individuals’ interests and skills can be business sensitive.
The cultural perspective relates to implicit, unwritten societal and cultural norms and expectations. For example, how can a society welcome the idea that algorithms would increasingly meddle with the social fabric and networks of people? This demands an understanding of how such systems can avoid creating confrontations between overly dissimilar people, particularly in unstable societies.
The ethical perspective relates to written rules, such as law and ethical regulations, particularly about the use of big social data for PSM. Building a suitable collection of social data is unthinkable, as governance and regulations on gathering and using it continue to be developed.
