I have subscriptions to many newsletters and when I recieve one in my inbox, I first scan it
too see if any of the topics interest me and if so, I click on the link and transfer to the web page to read its contents. Again, when I
reach the web page, I first scan it to discover the most interesting parts. This is very time consuming and tiresome.
As a solution, I propose a Virtual Agent which is infact one's virtual avatar in cyberspace. For a sufficient amount of time,
this agent monitors the user's habbits, interests, relations, entry and exit points in cyberspace. Then, it creates a very detailed and upto date profile of the user.
After the agent has been trained, it starts collecting and consolidating information that might likely be what the user would try to collect on his own. By a combination of text-to-speech and
speech-to-text technologies, user can direct the agent on particular topics by issuing voice commands in natural language. Also the agent can vocalize the information it
has gathered so far to the user.
It will also be possible to rate the agent based on the relevance of the information it has supplied either by manual or automatic means.
Virtual Agent can re-train itself continously by monitoring the user's reactions to the information it has supplied.