Whilst I usually find people are either bored or overwhelmed by explanations of my work at university, this is a question I’m frequently faced with at the moment. In Computer Science, our equivalent of a dissertation is researching, designing and developing a system for a particular purpose; the ‘Final Year Project’. Fortunately, the principle behind my project is, in fact, quite relatable and easy to understand.
Anyone who uses email will have been subscribed to a mailing list, e.g. a company newsletter, and therefore familiar with receiving the junk mail that companies send out to their customers, most of which is of absolutely no interest to us. Personally, when I start getting messages like this I tend to unsubscribe from the mailing list and I’m sure many of you do the same, or perhaps just delete them straight away. This is a problem not just for us as customers, but also for the companies sending out the mail; we don’t want to be hassled with emails that don’t interest us, but unsubscribing means that we miss out on the few messages that were of interest. The company also misses out on potential custom from those messages, so it is in everyone’s interest to solve the problem. My project is to develop a system which gives the users more control over the mail they receive and ultimately only sends the customer messages that interest them. It also removes any need for the company to guess who might be interested in the message when targeting customers.
This is approached by placing the system between the company and the customer. The company sends the email to the system, the system analyses the message to work out which customers will be interested and then sends it those people only. The primary problem I’m solving is how the system makes the decision as to who will be interested. Essentially, it is an artificial intelligence problem and is based on extracting keywords from emails to compare them with keywords which interest the customer. Hence, the official title of my project is ‘Learning Keywords in Email Messages’.
So there you have it! That’s the extent of the detail I’ll go into here but perhaps I’ll write a more technical post a little further down the line.