CEAS 2006
Brainstorming Session on “What's missing?”
7/28/2006
Overlooked or substantially more research needed
Search engines for e-mail archives (e.g., query expansion, hubs/authorities).
N-factor performance evaluation.
Reputation vs. privacy.
Emergence of rep as requirement; impact on privacy such as making anonymity impossible?
Real time usefulness of analysis techniques.
Researchers need to consider computational load and other performance impact on real-time analysis.
How many spammer are there?
Policy people need this info (cf, Linford) because it has a large effect on how to approach dealing with the problem
Cell phone spam/text messaging.
Emerging problem
Image spam
Taking research from efficient computer vision algorithms and applying them to spam detection
Detecting forwarding and handling spam in forwarding
research, standards, proposals for standardisation of forwarding
Dealing with Advanced CSS based SPAM - advanced html spam
If spammers become advanced, can hire cheaply programmers from countries to which we outsource software, they can become really advanced in what they can send in spam
Research on how to tackle the new variants of spam and a common technique to deal with any new variants of html obfuscation.
Collaborative detection of botnets by anti spam communities
Research on how we can help the government by providing thru a secure
backchannel, information and
evidence about spammers
Collaborative Detection of botnets on internet
Browser/operating system mandatory enhancement to monitor port 25 connections and reporting users about arbitrary port 25 connections - hence detecting botnets
Methods from other scientific areas
Usability evaluations using HCI methods, not just accuracy figures.
Learning model for adversarial classification.
Remove human factor from decision-making
Lack unified model; need one
Economic restrictions on the usefulness of spam – is there a mutation rate that is so high that spam is no longer profitable?
Economics and security; need to cross-pollinate the two areas, to work together; spam as a problem is a consumption of our time/resources
Can better filters make spam uneconomical
Lessons from war
Can the methods used to study military tactics help us in the fight against spam? Possible collaborative topic for spam researcher to work with people from the national war college or other academic military strategy field.
HCI and psych of users interacting with junkmail filters
Do they feel like they are ‘combatting’ spam?
“User perceptions” of the mechanisms.
New products or services that require novel research
Search for email from dead people
Social network & content techniques needed
Need corpora that are relevant to this r&D
Importance of layout (Format templates)
Eg, “Mortgage” spam.
Look at web, to find email spammers
Eg, pharmacy spam leads to web links, and web pages produce common terminology that makes identifying spam email easier.
VOIP spam; real-time spam; eg, based on noticing current activity(!)
Authentication requires Accred/Cert/Rep mechanisms
Which mechanisms are the most useful?
What are differential aspects of using each?
How to bootstrap their use?
More realistic models of costs for delivery/non-delivery
Real-time performance issues. Collaborative filtering role? Requirement?
Domain names and phone numbers are cheaper; encourages more spam
R&D to hold registrars held (more) responsible
Partial user feedback
Partial correction; impact of imperfect performance on correcting
Other
Less on spam, more on anti-virus, time-delay for propogation, how companies use attachments, General email research.
More effort on lawsuits against spammers; research on impact
Moving from UBE definition to “unwanted”
How to classify; how handled; ...