Citation Context

...r example, CrowdSearcher (Bozzon et al., 2012)senables users to explicitly federate search engine queries tossubsets of their online social network, and SearchBuddiess(Hecht et al., 2012) and Cognos (=-=Ghosh et al., 2012-=-) suggest contacts who might have the expertise to answer asquestion. Q&A tools like Aardvark (Horowitz and Kamvar,s2010) and IM-an-Expert (White et al., 2011) use expertisesinformation to route quest...

Citation Context

...s (covering popular, niche, local and global topics) to avoid being restrictive and falling prey to the long standing expert versus crowd debate [6]. We leverage a technique that we recently proposed =-=[7,13]-=- to crowd-source ex-pert user detection in Twitter, to identify over half a million experts on a diverse set of topics of interest to Twitter users. We gathered two samples of tweets – (i) the 1% ran...

Citation Context

...s 2011). Overall, expert detection in Twitter is a difficult task. Studies show that tweets content provides less useful information than contextual data (profile, user list, etc.) (Liao et al. 2012; =-=Ghosh et al. 2012-=-). On the other hand, manual expert detection revealed that decisions are influenced by shallow factors such as author names (Pal and Counts 2011). In this work, we therefore do not attempt to identif...

Citation Context

...he method in [15]) to the original news article. We use standard evaluation metrics such precision, recall, and AUC, all measured after ten-fold cross validation. 5.2 Features Previous work including =-=[10, 24, 26]-=- provides us with some useful information about suitable features for the detection of curators. These include network-based features such as the number of followers of a user – shown not to be suffic...

Citation Context

...isit these two foundational questions about Twitter. Recently a lot of research has focused onmining user data to develop detailed characterizations of Twitter users. For instance, in our prior works =-=[12, 25]-=-, we have proposed techniques to infer the biographical information and topical expertise of millions of Twitter users. These recent advances in accurately profiling large numbers of Twitter users ena...

Citation Context

..., u’s followings). The topical expertise of u’s followings are, in turn, inferred from social annotations (collected via the Twitter Lists feature [1]) using a methodology developed in our prior work =-=[4, 10]-=-. We compare the topics of interest inferred for a given user by our methodology with those extracted by a state-ofthe-art technique – using supervised topic models (Labeled LDA) on the tweets that ar...

Citation Context

...e-scale information sharing platforms. Indeed, many efforts have focused on finding experts in online forums [29], question-answering sites [18], enterprise corpora [3, 5], and online social networks =-=[8, 11, 22, 25, 30]-=-. These approaches, however, have typically focused on finding general topic experts, rather than local experts. In this paper, we investigate new approaches for mining local expertise from social med...

Citation Context

... of the members into his own list. Any combination of following, listing and subscribing is also possible. While there is preliminary research on the use of lists to infer latent user characteristics =-=[5, 8, 9, 11]-=-, surprisingly there is little research so far on how these three forms of information consumption are related to each other. An examination of lists will not only help us understand information consu...

Citation Context

...2005 TREC Enterprise track show that user expertise can effectively be derived from email content; the persons being cc’d in an email were often authorities on the content of the message. Ghosh et al =-=[10]-=- leveraged social media (Twitter) content for seeking experts on a topic. Their results indicate that endorsement in other users’ Twitter Lists (of which the topics need to be extracted) infers a user...

Citation Context

...he topical expertise of the Twitter users who have posted it. To obtain the information about users’ topical expertise, we leverage a methodology based on Twittre Lists, developed in our prior works (=-=Ghosh et al. 2012-=-; Sharma et al. 2012). We show that our methodology performs better at inferring topics for posts than a state-of-theart publicly deployed commercial topic inference system. Our study conducted using ...