Hi, Does anybody know any method(or research done) to detect a fake user in a website like Twitter or LinkedIn; by "fake user" I mean an account which its user is not a human being and can be a kind of a bot that is trying to harvest the data of a specific user or a group of users; many websites like Facebook are very popular; many non-democratic governments want to spy on their people with whatever means possible, so I think it will be an interesting question be asked that "are users of a website like Facebook safe against automated harvesting of their data by a bot?".
In financial reports some income tax authorities use deviations of the digits distribution from Benford's law to detect forgery. The statistics of social networks is similar to that of Benford's law namely, f(n)=ln(1+1/n)/ln(N+1) ,where N is the number of the users per given time of a given website. It will be interesting to check this formula and to see if a user does not follow this statistics.
Kafri, O. (2016) “A Novel Approach to Probability”, Advances in Pure Mathematics, 6, 201-211. doi:10.4236/apm.2016.64017.
Dear Hassan Abedi,
please check Section 4.3 of http://arxiv.org/abs/1504.05655 (or in the attached link), for a list of previous work in that subject.
There are a series of works already published in this topic, including one that uses Benford's law as strategy for anomaly detection, as suggested by Mr. Oded Kafri.
Article Online Social Network Analysis: A Survey of Research Applica...
If you are happy to check one at a time, then the easiest way is to check the profile picture. A fake account will probably have taken a picture off the internet. All you have to do is search for that same picture using google images. If it comes up next to something totally unrelated to the claimed identity, it is probably a fake profile. do the following:
Click and drag their profile picture to your desktop.
In a new tab, open google images.
Drag and drop the profile pic into the search field: it will expand.
Google will either return an exact match (with information like names), or pictures similar to the original.
If it returns exact matches, then you can visit the other sites to see who it relates to. You may find, for example, a linkedin profile or other facebook account using the same profile pic.
I do not have a precise reference in mind, but I saw several works on this topic, if it can help you:
-Some people look at contents of publications, if available. Spammer are easy to spot this way
-Some people look at temporal patterns of activity. Human behaviour follow some typical patterns (burstiness of activity, sleeping hours, etc...)
-If you have none of these, you can still use the network of friends/followers/followees. I saw simple methods proposed (number of contacts, clustering coefficient among friends, etc...) but some people have proposed more advanced one (based on advanced network topology analysis). Search with keyword "link farmers" for instance
Sorry I do not have direct link to publications, I just checked quickly that some of these papers were published in WWW conference 2013. You can start your research from there ! ;)
Hi Hassan,
Remy is right, you can consider publication in ICWSM or ASONAM conference too which may often be related to this kind of topic.
You can check papers of Kyumin Lee et al. about spams and crowdturfing that are truly great (e.g. " Who will retweet this?: Automatically Identifying and Engaging Strangers on Twitter to Spread Information " or "Detecting collective attention spam"), the papers of Fabricio Benevuto about bots ("An empirical study of socialbot infiltration strategies in the Twitter social network") or my papers about specific Twitter users called Social capitalists (kind of spammers) : "On the importance of considering social capitalism when measuring influence on Twitter" or "A reliable and evolutive web application to detect social capitalists".
Nicolas
Hi Hassan,
some time ago I've created about 3 thousand Facebook test users. In general, fake users lack some common functionalities of real accounts. As an example it is not possible to request for friendship.
Paolo
Hi @Paolo Pilloni
Could Facebook effectively detect your fake accounts?, could you elaborate a little, because I'm wondering how good Facebook can protect its users' data against governments(non democratic) spying on their people. I think it's a serious matter that many users must be informed about; and thanks in advance.
There are many research papers about this. Here are a few examples
Article Uncovering Social Network Sybils in the Wild
Article Detecting spammers on social Networks
Technical Report BotOrNot: A System to Evaluate Social Bots
Article Íntegro: Leveraging Victim Prediction for Robust Fake Accoun...
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining