In my thesis, we deal with the recommendations. Personalized recommendations are widely used and very popular. However, one of the basic problems is the distrust of users of recommendation systems. They consider them as intrusion of their privacy. To solve these problems, we decided to focus on explanation of each recommendation item to the end user. In this context, we have created a recommendation system ExplORe, that uses collaborative filtering as a standard recommendation technique for recommending news articles. Under this system, we have designed and implemented our hybrid method of personalized explanation of recommendations. We also performed several experiments in order to verify our method.
One part of our solution was the UX study with eye-tracking tool. We need to perform such a study in order to obtain basic information about the visualization of explanations (e.g. the number and location of explanations). Subsequently, we use these information to design our system Explore.
Actually, this was my first UX study so I did not know what it will be like. First of all I created a few designs of system user interface with focus on explanations. I ask some friends to come at my school and be part of science experiment. Yes it looks really interesting and worthy of the time when you say it this way :). Fortunately I found a few willing people. I would like to thank them this way.
After I have a bunch of people I needed to create personalized designs of news articles for each person. I decided to make it with graphic tool instead of create a first simple draft of web page. Ok that was not the smartest decision. First of many wrong decisions from me. I found that as wrong after I created first twenty designs. But these twenty was just for one person. Can you imagine this? I can, but I still do not know why I continued to do it like this and not create simple html page for this purpose after that first twenty designs.
Ok shame on me but after all I did it. After that I created a time schedule for every participant of experiment. I think that I can make one person in like 20 minutes. That was my second wrong decision. I needed around 30 – 40 minute for one person. One and only good thing was that nobody needs UX room right after me.
Next week, I’ll start to evaluate the results of experiment. Again, I expect that I can do it in 4 hours. It can not be so much time consuming because I just need a few tables. Ok right now you definitely know what is coming. Another WRONG. I needed around 8 – 10 hours. You know, I think that I will need every metrics that Tobii studio provides. Actually, I barely used 4 of them.
Finally, at the end of these two weeks I did the experiment with every participant and also evaluated it. My advice for everyone who wants to do their first UX experiment. Try to double your time estimations. But at the end of it, I think that every participant of experiment like it and enjoyed it just like me. But maybe it is just another wrong thought. I have some experience with that :).