RESEARCH & DEVELOPMENT
After the decision of spread Caronte Antispam as distributed application, we have created a small department of research and development .
The members of this department, are the same that have conceived and developed Caronte Antispam.
You should not think of their as a formation of programmers with business procedures and marketing rules, but as people that take care of the spam problem and all that concern it, trying of create a path of study and development.
Caronte Antispam TEAM
The artificial neural network and its concepts based on our brain architecture, are today real software applications and are used a lot in medical and statistic field.
What interests us in this contest is certainly the statistic field and apply a mechanism of “Neural Network” to spam is certainly a bet and a look to the future.
A neural network can exit from all the behaviours of “boolean” logic and give statistics answers to complex issues, that would be impossible to analize only with a logic of “true” and “false”.
“Anna” in its prototype is already in yard and in test phase, and it was relatively easy do it's first LAYER, just think how much fit all behaviors that are created by spam on logical analysis.
The concepts of statistic analysis implemented for the moment in ANNA are :
-Analysis of behaviours on the SMPT comunication protocol maked from client that sends email.
- Monitoring of the entire system antispam with relative intelligent messages of alert .
- Autoconfiguration of antispam systems while in load phase.
- Share of the experiences of collective memory based on the users of your antispam server.
For us “Anna” is already a tangible and running mechanism, there are all the issues of learning of the network still to be resolved. As it is right in the DNA of the artificial neural network the experience induced from system of learning to make that everything works, and being every server or antispam system “unique”, fundamental it's to make a learning system that adapts to the requirements of the user that has not so familiar with these concepts.
For be more explanatory, we can't give you “Anna” with base concepts already learned by default , because the unique way to learn such concepts, is that “Anna” understands where it is installed and what it has to manage. If default teach that 500.000 daily messages is a load to manage while 200.000 is the normal, on others realty such experiences aren't good with the relative consequently that Anna will not work.
Indeed when we apply Anna to Caronte Antispam we resolved the learning issues of the network and we distribute to all of you such resource.
Officially is not decided in what release of Caronte Antispam, Anna will be inserted as resource to be used, but being a system out of all the schemes, it will need indeed of a long period of test.