RACOFI Music is being replaced by inDiscover. Please email Sean.McGrath@unb.ca for details.
At the most basic level, RACOFI is a rating site like many others (e-pinions, Yahoo Music and so on). However, its focus is on offering a "personalized recommendations" to the users. As of now, the web site focuses on Canadian Music, but then intent is to use Music as a model of learning objects. It is based on years of research on rule-based systems and collaborative filtering algorithms. We believe that at the conceptual level, this web site is a step beyond anything else on the web right now.
Open a web browser and go to http://racofi.elg.ca If you already have a user name and password, enter them into the correct fields and click on the sign in button. If this is your first time using RACOFI Music click on the "here" link following "If you are a new user..." Then pick a user name and password and enter them into the correct fields and click on the submit details button.
After logging in you can now see a list of the top 10 highest rated songs within the system. This list is not personalized since you need to rate items before you can get personalized recommendations. Along the bottom of the screen is a menu of buttons that you will use to navigate through the RACOFI Music system. To begin rating items click on the "Browse Artists" button which is located on the left side of the navigation menu along the bottom of the screen.
You now have the option to rate music by artist or to rate randomly suggested albums. If you wish to rate an album from a specific artist click the first letter of their name, and then select the album you wish to rate. I f you wish to rate random albums click on the "here" link following "To rate random items..."
You can now see the page for an individual album. The average user ratings for this album are shown in a graph on the upper right side of the screen. Now select a value from 0 - 10 (with 0 being very poor and 10 being very good) for each of the 5 rating categories: impression, lyrics, music, originality and performance. After you have selected a value for each category, click the submit button. Continue to rate as many albums as you wish.
After rating several albums click on the "Recommendations" button on the navigation menu at the bottom of the screen and RACOFI Music will recommend albums that you might enjoy based on your preferences. Select which category you would like the results sorted by and select whether you would like to use the simple or complex algorithm. By changing algorithms you will change your results. You can then click on the search button and perform a more detailed search of the RACOFI Music system.
RACOFI Music (Rule Applying COllaborative FIltering) is a multi-dimensional music recommendation site, with an exclusive focus on Canadian content. RACOFI Music will suggest music albums that you can expect to enjoy, based on your rating style and on the rating style of others. Your rating style is discovered by having you rate albums that you like and dislike. The more albums you rate within the system, the more accurate RACOFI Music's predictions will be.
RACOFI Music also provides advanced searching capability which allows you to search for albums with specific criteria and to also specify a level of importance to the searching criteria. For example, you can search for an album by specifying that you really want it to have good music but you do not really care if the lyrics are high quality or not.
RACOFI Music is a specific implementation of the RACOFI system. RACOFI can be easily applied to many domains allowing you to use this new technology for more than just music.
RACOFI can save you time. You no longer have to sift through large amounts of data in order to find the items that interest you most.
RACOFI can provide you with exposure to new possibilities by suggesting items that you can expect to enjoy but might otherwise, have not considered on your own.
RACOFI's mandate , in part, is to promote the Canadian music scene. Artists can submit information on themselves (an informational URL is required) to be added to our databases.
RACOFI's philosophy is quite simple: to share with you items which other users, who are similar to you, have found to be valuable. RACOFI works by comparing your rating style to users who have similar rating styles by using collaborative filtering algorithms and then makes recommendations of items you are likely to enjoy.
RACOFI Music works by having you rate contemporary Canadian music. The collaborative filtering Pearson, STI nonpersonalized 2-step, and the Per Item Average algorithms are then employed together with averaging rules to recommend music objects that best match user queries. RACOFI allows you to search the system for items based on criteria that are important to you. When searching, RACOFI allows you to specify a "weight" or a level of importance on its searching criteria.
A bunch of researchers including Harold Boley, Daniel Lemire, Stephen Downes, Hélène Fournier, Rodrigue Savoie, Bruce Spencer, Jo Lumsden, and others. The server is maintained by Luc Belliveau who is a valuable resource. Many COOP students contributed to the project including Michelle Anderson, Nancy Howse, Stephen Greene, Marcel Ball, and Sean McGrath.
Please contact us if you are interested in this technology.
Yes, this is very easy. Here's an example in HTML.
1:<html> 2:<head> 3: 4:<script type="text/javascript" 5:src="topten.js"> 6:</script> 7: 8:</head> 9: 10:<body> 11: 12:<script type="text/javascript">writeTopTen();</script> 13: 14:</body> 15:</html>
You must also put a file called topten.js with the following content:
1:function writeTopTen() { 2: document.write("<script type=\"text/javascript\" src =\"http://racofi.elg.ca/servlet/TopTenServlet?javascript=true\">"); 3: document.write("</script>"); 4:}
Sean McGrath and Stephen Greene review and maintain the album database.
We have published the following related papers...
Also, Harold Boley gave a tutorial on RuleML which used RACOFI as an example at EPIA'03.
Ok, I've had enough information. Now I want to go back to RACOFI home page!