{"id":304418,"date":"2012-02-14T14:45:08","date_gmt":"2012-02-14T22:45:08","guid":{"rendered":"https:\/\/newed.any0.dpdns.org\/en-us\/research\/?p=304418"},"modified":"2016-12-16T12:40:45","modified_gmt":"2016-12-16T20:40:45","slug":"social-search-social-builds-bridges","status":"publish","type":"post","link":"https:\/\/newed.any0.dpdns.org\/en-us\/research\/blog\/social-search-social-builds-bridges\/","title":{"rendered":"Social Search Social Builds Bridges"},"content":{"rendered":"

By Janie Chang, Writer, Microsoft Research<\/em><\/p>\n

Two conferences, same region, consecutive weeks\u2014sounds like a perfect opportunity for a social.<\/p>\n

That\u2019s certainly what sprang to mind for Meredith Ringel Morris<\/a>, researcher in the Natural Interaction Research Group<\/a> at Microsoft Research Redmond<\/a> when she realized that the Association for Computing Machinery\u2019s (ACM\u2019s) fifth International Conference on Web Search and Data Mining (opens in new tab)<\/span><\/a> (WSDM 2012) would be occurring in Seattle from Feb. 8 to 12, the week before the ACM\u2019s Computer Supported Cooperative Work (opens in new tab)<\/span><\/a> conference (CSCW 2012), to be held Feb. 11-15 in nearby Bellevue, Wash.<\/p>\n

\u201cSince the two conferences were happening in close proximity,\u201d Morris says, \u201cI thought it would be interesting to hold a bridge event over the weekend to bring the two communities together.\u201d<\/p>\n

The result was the Social Search Social, sponsored by Microsoft Research Connections<\/a> and held on Microsoft\u2019s Redmond, Wash., campus on the evening of Feb. 11. Morris, though, had more than a cocktail party in mind. It was an opportunity to foster networking and collaboration between the conferences\u2019 research communities.<\/p>\n

Combining Different Perspectives<\/h2>\n

\"collaboration\"The number of researchers who belong to both communities is small. Morris\u2019 goal was to provide a networking opportunity for both groups, because there is a great deal of affinity between them. She believes that if the two communities were to work together more often, they could achieve interesting research advances.<\/p>\n

\u201cThe WSDM community is very focused on data mining, machine learning, and statistical analyses,\u201d Morris says, \u201cwhereas researchers in the CSCW community are interested in user-interface innovations and how users interact with technology. Although they approach social search from two different perspectives, it\u2019s an area relevant to both disciplines. Combining these different perspectives will be critical to developing next-generation social-search technology.\u201d<\/p>\n

Morris points to two papers as examples of synergy between the two disciplines. The first is Tweeting is Believing? Understanding Microblog Credibility Perceptions<\/em><\/a>, co-authored by Morris and her Microsoft Research Redmond colleagues Scott Counts<\/a>, Asta Roseway<\/a>, and Aaron Hoff, along with Julia Schwarz of Carnegie Mellon University. The paper was nominated for CSCW 2012\u2019s best-paper award.<\/p>\n

The other is the WSDM 2012 paper How User Behavior is Related to Social Affinity<\/em><\/a>,<\/em> co-authored by Rina Panigrahy, Marc Najork, and Yinglian Xie of Microsoft Research Silicon Valley.<\/p>\n

What Influences Credibility<\/h2>\n

Traditionally, people place more trust in content that originates from people in their social network.<\/p>\n

\u201cIf you follow someone because you know them personally or by reputation,\u201d Morris says, \u201cyou tend to view their tweets as credible. But when you search on tweets, you need to assess the trustworthiness of people outside your network.\u201d<\/p>\n

\"Meredith

Meredith Ringel Morris delivers opening remarks during the Social Search Social. (Photo by Jeremy Pickens)<\/p><\/div>\n

Morris and her team conducted experiments in which they systematically manipulated several features of tweets\u2014such as Twitter handle, avatar appearance, and message topic\u2014to assess how these factors influenced participants\u2019 views of the credibility of tweet content.<\/p>\n

\u201cWe had to structure an experiment that used both factual and false tweets,\u201d Morris recalls. \u201cThe false tweets, however, had to be plausible, because anything that was obviously untrue wouldn\u2019t really test perceptions of credibility. It was amazingly difficult coming up with false but plausible content.\u201d<\/p>\n

There\u2019s more at stake than mere celebrity gossip. During political campaigns, people have been known to tweet false information about candidates\u2014something that could influence voting decisions. In health care, misinformation can result in harmful decisions.<\/p>\n

The researchers found that users are poor judges of truthfulness based on content alone; instead they are influenced by heuristics such as whether the tweeter has a topically relevant user name when making credibility assessments.<\/p>\n

\u201cAs a result, we have some suggestions for how people might read tweets with a more cautious eye,\u201d Morris says, \u201cbut we also have some useful ideas for how social search pages might redesign their user interfaces to present information that could help people make more informed judgments.\u201d<\/p>\n

More informed search capabilities also can help users make more informed judgments. Panigrahy and his colleagues are studying ways to define measurements that represent relationships in a social network.<\/p>\n

When measuring proximity in a social network, the shortest-path distance between nodes is an obvious candidate, but shortest path by itself neglects to take into account the number of connections shared by two users. How User Behavior is Related to Social Affinity<\/em> introduces a new measure of \u201caffinity\u201d in social networks that takes into account not only the distance between two users, but also the number of connections between them: the \u201crobustness\u201d of their connection. The team found that users who are close to each other\u2014in terms of either distance or affinity\u2014have a higher similarity in terms of demographics, queries, and tweets.<\/p>\n

Accelerated Networking<\/h2>\n

More than a hundred researchers from more than a dozen countries attended the Social Search Social. Researchers from universities such as the Massachusetts Institute of Technology, Stanford, Carnegie Mellon, Harvard, and the University of Washington mixed with industry peers from Microsoft, Yahoo!, Google, Amazon, Boeing, IBM, and LinkedIn. Most significantly for Morris, the guest list was split evenly between CSCW and WSDM attendees, evidence that each group was interested in the other\u2019s perspectives.<\/p>\n

Based on the nonstop discussions that transpired in every corner of the room, the Social\u2019s success as an icebreaker exceeded even the expectations of co-organizers Morris, Gene Golovchinsky, Jeremy Pickens, Chirag Shah, Michael Twidale, and Madhu Reddy. The main activities were:<\/p>\n