<html><head><style type='text/css'>p { margin: 0; }</style></head><body><div style='font-family: arial,helvetica,sans-serif; font-size: 10pt; color: #000000'>(Resending to correct subject line)<br><br><hr id="zwchr"><div style="color:#000;font-weight:normal;font-style:normal;text-decoration:none;font-family:Helvetica,Arial,sans-serif;font-size:12pt;"><style>p { margin: 0; }</style><div style="font-family: arial,helvetica,sans-serif; font-size: 10pt; color: #000000">Arpan Ghosh will present his MSE thesis talk on Wednesday May 1 at 3PM in Room 301 <br>(note room!). the members of his committee are: Mung Chiang, ELE (advisor) and <br>Andrea LaPaugh (reader). Everyone is invited to attend his talk. His abstract follows <br>below.<br>------------------------------------<br><br>Abstract:<br>Tracking the Business Pulse Using Twitter<br>While Twitter surged to popularity soon after its creation, it took a few years to find legitimate & profitable uses for it. <br>Just like it brought fans closer than ever to celebrities, people closer to real-time news and voters closer to presidential <br>candidates, today it is helping define a new paradigm for consumer businesses to interact with their past, present <br>and future customers. Twitter makes it very easy for businesses to directly ‘reach-out’ to the customers that follow <br>them, for the purposes of announcements, promotions and advertising. The return path of the communication loop <br>wherein a company would like to analyze customer needs, sentiment, feedback, reviews and grievances, is much <br>harder for several reasons. Twitter as a data stream is very sparse and has a low signal to noise ratio as it continues <br>to be dominated by tweets that are basically about nothing: personal conversations, random observations or spam. <br>It is a slight exaggeration to classify tweets as semi-structured data, even with the presence of hashtags, due to the <br>dynamic method of Hashtag creation and their longevity. Twitter also provides very sparse information about the users <br>authoring tweets. This work is driven by the vision of building a tool that allows businesses to monitor, analyze and <br>gain insights, in near real-time, of the opinion, sentiment and feedback that exists about them on social networks like <br>twitter, analogous to what Google finance accomplishes for the financial aspect of businesses. To this end we make <br>the following contributions: 1) A quantitative analysis of twitter usage to determine its suitability for this business-oriented <br>use case. 2) A classifier filter out ‘irrelevant’ tweets for the aforementioned use-case. 3) An algorithm to detect trending <br>increased activity on twitter pertaining to a business and cluster periods of high activity into logically separable stories/events<br> in the business’ timeline. 4) An algorithm to classify people talking about a business on twitter based on their <br>expertise/influence in the area/topic that the business specializes in.<br><br><br></div><br>_______________________________________________<br>talks mailing list<br>talks@lists.cs.princeton.edu<br>https://lists.cs.princeton.edu/mailman/listinfo/talks<br></div><br></div></body></html>