CSCI8380 (Spring 2012): Paper Review Form Reviewer Name: Anuj Shetye Paper Name: Information creditability on Twitter Section I. Overview A. Reader Interest 1. Which category describes this manuscript? ___Practice/Application/Case Study/Experience Report _X__Research/Technology ___Survey/Tutorial/How-To B. Content The authors of this paper focus on automated methods for information creditability of news propagated on twitter. The authors classify tweets credible or not by specifically analyzing the micro blog postings related to recent topics based on features extracted from them. C. Presentation 1. Does the introduction state the objectives of the manuscript in terms that encourage the reader to read on? _X__Yes ___Could be improved ___No 2. How would you rate the organization of the manuscript? Is it focused? Is the length appropriate for the topic? _X__Satisfactory ___Could be improved ___Poor 3. Please rate and comment on the readability of this manuscript. ___Easy to read _X__Readable - but requires some effort to understand ___Difficult to read and understand ___Unreadable Section II. Evaluation Please rate the manuscript. Explain your choice. ___Award Quality ___Excellent _X__Good ___Fair ___Poor Section III. Detailed Comments (provide your thoughts/criticism about the ideas in the paper; not only summarize the paper but have a critical look here) The paper talks about creditability of information on twitter, a popular micro blogging website. The authors analyze tweets on micro blogs and classify them as credible or not credible based on some features extracted from them. The researchers focus on the feature extraction like no. of followers, age and the url to the source website using Amazon mechanical turks to analyze the information. The methods use J48 decision tree algorithm which is good but should have been described more. The evaluations are done by human assessment rather than employing an automated method. Additional Comments: 1. Provide one aspect that you liked the most in this paper. The use of J48 decision trees The authors did extensive evaluations on their proposed method. 2. Provide one aspect that you disliked the most in this paper. They rely mostly on human assessments rather than trying to automate this stuff to. Section IV. Discussion Points (provide at least 3 discussion topics/questions related to ideas/techniques described in the paper; these will be used for discussions in the class) 1. The J48 decision tree has the tweet with url at the root node, but how do we decide that the URL itself is credible ? 2. Information credibility will matter most when it is done at real time. The paper talks about tweets after a certain amount of time has passed.