Do You Know? Recommending People to Invite into your Social Network (IUI-2009 Assigned Reading)

Do You Know? Recommending People to Invite into your Social Network (IUI-2009 Assigned Reading)

By: Ido Guy, Inbal Ronen, Eric Wilcox

Summary:

This paper talks about a UI and system that provides people with a set of recommendations of people it thinks match the profile for being a candidate to be in their social network. This system is based on an aggregated information set about peoples relationships which is retrieved using SONAR. SONAR is a system for collecting and aggregating social network information across the organization. SONAR extracts an employee’s social network by getting the information about relationships between people. The relationships include information about: organizational charts, paper co-authorships, patent co-authorship, direct connection as well as several other things associated with IBM. The UI allows for the scrolling through of recommended people one by one while it shows the “relationship evidences” as to why that use was selected as a candidate. People are recommended based off of an articulated set of social network information. This interface is called the “Do You Know” widget and is a new addition to the homepage of IBM’s next-gen employee directory. This recommender system shows a profile of the person as well as a list of things that are in common between the user and candidate. The DYK application recommends people who you may be familiar with, but are not yet connected with.

The evaluation consisted of 2 parts. The first was a field study where the use of DYK was monitored for 4 months. The second was a qualitative user study that includes things such as interviews, and surveys.

There were some very important discoveries with the usage of the DYK system. Many users said that even though the information about links between them and the candidate were known, that they were still useful because it invoked trust in the application. They also found out that even though users were given over 100 results at times, that users of DYK rarely searched through more than the first 20.

This evaluation showed that people recommendations can be highly effective in increasing the number of connects between users as well as the overall number of users in a Social Network Site. Greater ease in finding potential connections was found to enhance user utilization of the site as well.

Discussion:

This paper seemed very familiar to an expert recommender system paper that I read earlier in the semester. As I said then I think this is an interesting area of study. To be able to access data from people and then connect links between them is a very important aspect of life. It is important to be able to contact people when needed especially for work. With the advancements in the Social Networking System I feel that we are going to see many advancements in this area. I thought this paper was a little drawn out, but for the most part well written.