I enjoy writing code. I have developed the tools I use in my research and I have created some other products I think are useful. The problem I have is keeping up.
Back in 2006, I took an open source, social bookmarking product and added my own modifications. I wanted to create a service that prioritized search suggestions. So, I extended the social bookmarking product by adding a system that counted the number of times suggested links were followed and also allowed users to rate the value of sites they visited. These data were saved and search results were ranked by a combination of these data. My little project continues to be used by those who find it and I was even asked for my code.
Time passes and technology projects age. The original Scuttle was upgraded to Scuttle Plus. I have no idea if any of my code ended up in this project. One of the problems with making your own modifications, especially to projects that involve a database, is that you cannot easily take advantage of upgrades to the “official” project. When improvements have been made by others, merging your modifications with an improved product means you have to carefully go through the new product and understand how the code has changed, determine if the modifications you made can be inserted in the same way, and consider whether changes in database structure require some other modifications. To tell the truth, I forget where in the code my changes even are. So, I am stuck with my 2006 era code.
One thing I came to realize is that my ranking scheme is confounded with time. Some records have been in this database for four years and have accumulated more hits (on average) simply because they have been available longer. This confounding increases in severity because older records are then displayed first and the suggested links followed more frequently. This results in suggestions that may not be suited to current interests. The ratings feature worked great and could potentially counter balance the time confounding problem, but I had to protect the opportunities to rate sites behind a password system. Scuttle was fairly open as I guess a social bookmarking site should be and I began to encounter trouble makers who added inappropriate content. Can’t have that if you hope to offer a service to educators. By limiting inputs behind a password system, you seem to discourage raters. Works great, but too much trouble to use.
Counting the number of times a link to a recommended site is followed does not create security problems and I am attempting to generate a system that would weight popularity by time. I suppose I could create a variable that continually divides the number of hits by the number of years a link has been available. There must be a way to make use of the information in the tens of thousands of times links have been explored from this site.
Maybe I will just wait until summer and then take a look at modifying the official upgrade.
