The Foafing the music

Site URL: http://foafing-the-music.iua.upf.edu

The system has shown the following fatal error when I login the system :

Fatal error: Cannot redeclare array_intersect_key() in /usr/local/share/foafing-the-music/include/General/functions.inc.php on line 26

I have tried to contact them, but I haven't got any reply from them. It seems they don't manage the site these days. If anybody know the current statues of the site, please let me know it. I will appreciate your help.

BTW, I read a paper (FOAFING THE MUSIC: A music recommendation system based on rss feeds and user preferences) about this site. In this paper, the author explains how the music recommendations are generated in the system. At first the system get interests from user's FOAF profile and detect artists and bands. And then, the system select related artists, from artists encountered in the user's FOAF profile and rate results by relevance. The following figure is the recommended artists from artists detected in a user's FOAF profile.
To get the artist's similarities, a focused web crawled has been implemented to look for relationships between artists (Such as: related to, influenced by, followers of, etc) . If the system detects the following artists from the user’s profile: Dogs d’Amour, Social Distortion, The Misfits and The Pogues. Starting from these artists, the system searches for similar artists and for artists influenced by them. Then, it scores them in terms of counting artist occurrences.

A Trust-enhanced Recommender System application: Moleskiing

Source type: Paper (Proceedings of the 2005 ACM symposium on Applied computing)
Source URL: http://portal.acm.org/ft_gateway.cfm?id=1067036&type=pdf&coll=GUIDE&dl=GUIDE&CFID=23661867&CFTOKEN=34778420
Summary:
The author introduce a real world application, namely moleskiing.it, in which provides personalized recommendation based on a blog oriented architecture which collects user experiences on ski mountaineering and their opinions on other users. The goal of this system is to make ski mountaineering trips safer by letting user report current snow conditions of ski routes and presenting to every user only information entered by reliable users. Using trust metrics, it allows to present only relevant and reliable information according to the user's personal point of view of other authors trustworthiness. In addition, the system provides open information exchange architecture using Semantic Web formats. It guarantees interoperability among different ski mountaineering communities.

Related site: http://www.moleskiing.it/mski/home.do

Using Semantic Web formats it provides open information exchange architecture.

The Foafing the music

Site URL: http://foafing-the-music.iua.upf.edu

The site is showing the following fatal error message when I login the site.



I reported the error message to the administrator of the site. Until they fix the problem, it is difficult to use the foafing music site.
One new thing in my blog, I received a comment from a foafing music blogger and he recommend a profile based recommend functions of the foafing music site.

'I Didn't Buy it for Myself" Privacy and Ecommerce Personalization

Source type: Paper
Paper URL: http://www.google.com/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Florrie.cranor.org%2Fpubs%2Fwpes03.pdf&ei=Yf30R-C4J5LYeYm53NoL&usg=AFQjCNEmEEP12VNt85dCNCB2wVPf5CTrlA&sig2=Q85IFKSdxhrTW5180wVlMA

The author, Lorrie Faith Cranor, is one of famous person in privacy research area. She is one of authors of P3P and APPEL Specification. In this paper, she outlines the privacy risks associated with personalization and describes a number of approaches to personalization system design that can reduce the privacy risks. Privacy risks can be reduced when personalization system designs allows for pseudonymous interactions, client-side data stores, and task-based personalization. In addition, interface that allow users to control the collection and use of their profile information can further ease privacy concerns. Pseudonymous profiles are a good approach when personalization information need not be tied to personally identifiable information. Client-side profiles may are useful when personalization services can be performed on the client. Task-based personalization may be appropriate when knowledge of a user's historical profile does not significantly enhanced a personalization service. Interface that put users in control of the collection and use of their data as well as the types of personalization systems more privacy friendly.

The Foafing Music

Site: http://foafing-the-music.iua.upf.edu

I saved "Elvis Presley" as my favorite musician in my profile in www.blogger.com. The foafing music site recognize my favorite musician is "Elvis Presley" and recommend new music releases like below;

The recommend new music releases information retrieve from amazon.com and iTunes store. If I click the link, the page move the site that contains the original information. The new music releases information took over 30 second to retrieve the information and generate the recommend page.

ConTag: A Sematic Tag Recommendation System

Source Type: paper
Source URL: www.dfki.uni-kl.de/~sauermann/papers/adrian+2007a.pdf
This paper introduce the Contag approahch. It generates semantic tag recommendations for documents based on Semantic Web ontologies and Web 2.0 Services. They designed and implemented a process to normalize documents to RDF format, extract document topics using Web 2.0 services and finally match extracted topics to a Semantic Web ontology.
ConTag is based on a Semantic Tag Recommendation Process like below:


1. During the first step, Normalisation, the document’s content is tranformed to
RDF format to gain a fulltext description.
2. During the second step, Topic Extraction, topics are extracted by requesting
Web 2.0 services. This results in a topic map using SKOS vocabulary (Simple
Knowledge Organisation System)
3. The Alignment Generation is based on document classification methods.
For each topic in the topic map, several weighted alignment possibilities are
computed to retrieve similar things.
4. The forth step is called Alignment Execution. The alignment scheme is visualized as tag recommendations. The user decides whether to accept or reject
recommendations.

Related sites
-Extracting relevant keypahrases
http://tagthe.net
http://www.topicalizer.com
http://www.dfki.uni- kl.de/~horak/2006/contag
http://phaselibs.opendfki.de/wiki/AlignmentOntology

The Foafing music

Site: http://foafing-the-music.iua.upf.edu

The foafingmusic retrieve the my top 10 most listened artists from last.fm like below;

From my listening habits, the service detect three artists (Jessica Simpson, Radiohead, Rhapsody). Using there three artists information, the system recommend related artists,
a review related with Rhapsody which linked to rateyourmusic.com, MP3-blogs that include the Radiohead's music.
They find out the recommend data by filtering the three artists names from their RSS database.

A Privacy-preserving Collaborative Filtering Scheme with Two-way Communication

Source Type: Research paper
Source URL: http://portal.acm.org/ft_gateway.cfm?id=1134742&type=pdf&coll=GUIDE&dl=GUIDE&CFID=20955699&CFTOKEN=40773225

Summary: In traditional CF systems, a server first collects ratings from users and then executes CF algorithms to make recommendation. There is a serious threat to individual privacy since data collected from users cover personal information about places and things they do, watch, and purchases. To solve this issue, a randomization approach has been proposed to disguise user ratings while still producing accurate recommendations. However, recent research work[1] has point out that randomization techniques might not preserve privacy as much as had been believed. This paper introduce a two-way communication privacy-preserving scheme in which users perturb their ratings for each item based on the server's guidance instead of using an item-invariant perturbation. According to their experiment, their new scheme preserve more privacy information than the randomization approach at the same accuracy level.

Reference:
1. Deriving Private Information from Randomized Data

The Foafingmusic2

Site URL: http://foafing-the-music.iua.upf.edu

To make my music list, I have visited last.fm several time and listed some musics. After that, I returned to the Foafing music site and updated the my music profile. The Foafing music site connected to the last.fm and retrieved the music list that I have listen in last.fm and generate the following information.

From the my listening habits, the system detected the two artists (Jessica Simpson, Rhapsody) and it provides related menu on left site.

The left menu provides information that related this two singer who was found in my listing list in last.fm.
I will explore left site menu next time.

An Effective Approach for Periodic Web Personalization

Resource type: paper
Resource url: http://portal.acm.org/citation.cfm?id=1248823.1249113&coll=&dl=

According to the paper, periodic web personalization aims to recommend the most relevant resources to a user during a specific time period by analyzing the periodic access patterns of the user from web usage logs. To support effective periodic web personalization, the author construct a user behavior model, called Personal Web Usage Lattice, from the web usage logs using the Fuzzy Formal Concept Analysis Technique. And then, the author deduces the resource that the user is most probably interested in during a given period. In the experimental, the author generates personalized resource for a set of predefined period conditions and measure the performance based on the applicability and satisfaction measures with respect to the different durations of period conditions and the different number of personalized resource (from 1 to 6). The experimental results shows that the proposed approach has achieved very effective web personalization evaluated by the applicability and satisfaction measures for predefined period conditions.

The Foafingmusic

Site URL: http://foafing-the-music.iua.upf.edu

The Foafing the music is a music recommender system based on user's profile

Based on your FOAF profile and your listening habits, foafing the music recommends the following things:

-similar artists to the ones you like
-new music releases from iTunes, Amazon, Yahoo, etc
-Album reviews from your artists and from recommended artists
-MP3-blogs to download music
-Postcast sessions to stream/download
-Automatic creation of playlist based on (only!) audio similarity
-Incoming concerts near to where you live!

When I create a new account, the system asked for my username or URL of three websites (Livejounal or Bloger, last.fm, and Webjay). From these three sites, they grab some music related information from my profiles.

Before I provide the URL of my blog profile page, I inputed "Dance, R&B, Latin" in favorite music and then saved the profile which is save in FOAF (the Friend of a Friend). From the FOAF profile in the blog site, the site get my favorite music information .

The site also request the user ID in Last.fm site. Last.fm builds a profile of my musical taste using a plugin for my media player. From this site the system gets noticed of music that I recently played.

At last, the site request the user ID in WebJay which is a tool that helps user listen to and publish web playlists. But currently, this service is closed by Yahoo.com

After I finish to create the account, the site took a time to retrieve my profile information from blogger.com site and last.fm site and showed my interests that retrieve from my profile.

The site provides three pages based on the my profile like below

- my artists page contains all the artists detected in my FOAF profile
- my music pages contains all the artists detected from my listening habits (from Audioscrobbler)
- my playlist page contains all the playlists that you have in Webjay (It is not working since the Webjay site is close)

In the artists page , the site showed that artist detected in my FOAF profile is "Dance". It seems the site recognize "Dance" as an artist name and showed one related artist "Shannon", so I added "Michael Jackson", "Elvis Presley" in my favorite music in blogger site and regenerated the my artists information in recommendation site. After that, the site showed that artists detected in my FOAF profile: Dance, Michel Jackson, Elvis Presley and provides following menu:
-related artists
-new releases
-reviews
-MP3-blogs
-Podcasts
-similar music

In the related artists menu, it show more than 30 related artists but I'm not sure what relationship are between them. I guess their music genre(Rock/pop) is same. If I click one of them, the site forward to MP3.com site and show the artist information.

New Contextual Advertising Solution

Resource type: news article
Resource url: http://www.centredaily.com/business/technology/story/393225.html

Openwave Systems Inc. announced the Openwave Contextual Advertising solution, a modular, end-to end advertising and content recommendation system that is designed to allow operators to capture value from targeted advertising by leveraging their existing subscriber relationships and unique network assets.
Their new personalization driven advertising solution reflects the multi-channel (mobile, PC and TV media) and user-based approach. This solution allows operators to offer subscribers the opportunity to receive only highly targeted and relevant advertisements, offers and recommendations, based on their immediate context, interests, location or demographic. Openwave's solution uses real-time context, user behavioral history, location information and voluntarily data to determine and recommend relevant ads and offers.

Openwave's modular Contextual Advertising solution including the following;

• Personalization system: it is a targeting engine that leverages key subscriber data
• historical behavior/transactions
• expressed preferences
• demographic information
• other data to infer user preferences
• Advertising broker
• delivery application modules
• ecosystem management
• advertising APIS

"About Openwave

Openwave Systems Inc. (Nasdaq:OPWV) is one of the world's leading innovators of software applications and infrastructure designed to enable revenue-generating, personalized services, including merchandising and advertising, which converge the mobile and broadband experience across all of a user's devices.

As the communications industry intersects with the Internet, Openwave software enables service providers to converge services, increasing the value of their networks by accelerating time to market and reducing the cost and complexity associated with new service deployment. Openwave's unique product portfolio provides a complete range of service management, messaging, location and client technologies. Openwave is a global company headquartered in Redwood City, California. For more information please visit www.openwave.com."

The exploration of Buy.com recommend system

I choose Buy.com which is my target recommend system that I'm going to explore during this semester. By interacting with the site, I will figure out how they reflect the knowledge and technique of recommend system into the site.

When I visited the Buy.com site, the site showed many items in main page to lead the customer's interest. The first curiosity is whether they provide different personalized main page to visitors or not. So, I visit the main page from two computers. From one computer I visited site anonymously and then I visited the main page of the site after I finished the login from another computer. The site showed the same main page to my anonymous visiting and authenticated visiting.
The site doesn't provide a personalized main page to their visitors.

In the main page, they showed many recommend items to visitor under the different subtitle like below;

-What's shakin': they shows the products with the highest change in sales rank over the past 24 hours.

- Deal of the week: they show one special items with a special price and free shopping. They compare the list price with our price, so it hook in customers.

- Featured Products: they show some featured products that are made by a specific company

- Most popular : they show popular items in their site.

- What's hot: it link to the What's shakin' page

- What's the buzz.. : they show the items that related with buzz at that time. Today buzz are tax, blu-Ray DVD player, Oscar nominees, grammys, and so on

- Featured promotions : they show all promotions items in this week.

- What there's more : they show the top-selling CDs from today's hottest artists at great low prices. They show how much customer can save money for each item.

This is all I have found recommend features in main page of Buy.com site. Next week, I will explore more deep pages in Buy.com site.

Target site: http://www.buy.com
Spend time: over 1 hours

A New Social Advertising Model

Source type: News article
URL:http://www.prweb.com/releases/2008/1/prweb663244.htm
The online marketing company TellAPal is launching a revolutionary system of advertising to enable advertisers to reach the hundreds of millions of users of social networks. Based on "referral marketing," TellAPal is the first recommendation based ad network optimized for social media and advertiser ROI. TellAPal's new system is completely user driven and it enables advertisers to empower their existing customers to recommend their product or service to their friends. Recommendations are made through personalized emails, IM, blogs, or social networking site and allow their friends to sign up for a special offer. If their friend sign up for the special offer, the recommender can earn rewards. This system shows how effectively advertisers can reach potential customers using recommend model

Simply say, this social adverting platform bring the word of mouth marketing, incentive marketing, and social media together and it works like below;

Step 1. An advertiser joins the TellAPal network and launches the program to their existing customers.

Step 2. Customers write a recommendation of the product or service combined with a special offer (ex. 30 day free trial).

Step 3. They user TellAPal's viral distribution platform to tell their friends.

Step 4. For each valid new customer, the referring customer gets a reward.

Resource: TellAPal.com

ATG gets Personal with CleverSet

Resource type: News Article
Resource URL: http://www.destinationcrm.com/articles/default.asp?ArticleID=7537

According to this new article, E-commerce solutions provider Art Technology Group announced plans to acquire recommendation service provider CleverSet for $10 million in cash. (Wow, CASH @_@) The acquisition aims to enhance ATG's existing personalization technology to create a more relevant online shopping experience, which will, in turn, increase conversion rates.

The CleverSet's technology is based on the theory of "the wisdom of the crowds" and it can obtains a more holistic understanding of what a consumer wants or may want. By incorporating this technology, E-commerce solutions can provide an automated system that takes the burden off the merchant and delivers the best product recommendation to their customers.

Referring to the Conneighton's reports, CleverSet customers have seen 20 percent to 50 percent lift in upsell, cross sell, and conversion rates. I think it is one of big reasons why ATG is going to acquire the CleverSet's technology.

John Lovett, senior analyst of site technologies and operations at JupiterResearch says, "My belief is that personalization will be best delivered not on a one-to-one basis, but more on a one-to-segment basis". Therefore, by leveraging the data of the individual and combining it with collective data, CleverSet can obtain a more holistic understanding of what a consumer wants or may want. I like his word "one-to-segment basis", it seem the personalization is more fine grained than before and close to the user's desire.

Autotags for music recommender systems

I found an interesting paper : "Automatic Generation of Social Tags for Music Recommendation"  that was presented at NIPS 2007.

This paper propose a method for predicting music social tags directly from music files. Using a set of boosted classifier, it map audio features onto social tags collected from Web. If there is a music DB with untagged or poorly tagged music files, this methods can make "autotags" for these music files and avoid the "cold-start problem" common in recommender system.