Socialfeed’s New Indicators

Long time no see! As I can spend only 20% of my time on this Socialfeed, unfortunately the blog always comes last.O Today, I would like to introduce 3 new indicators that we’ve added to Socialfeed to show our users what is happing inside the system.

1. Relevance
Relevance is the strength of the recommendation. This indicator shows how relevant each recommended item is to your interests as determined by Socialfeed. Socialfeed recommendations are based on the activities of a community of users with similar interests. The greater the number of people who have viewed the item and have interests similar to yours, the higher the Socialfeed relevance rating.

2. Influence
Influence shows you how much each user contributed to a particular item’s recommendation. In Socialfeed, recommendations are the result of the activities of people who are similar to you. Influence shows you whose activities are behind each recommended item and how much influence they had on the recommendation.

The orange bars are relative. You can see who has the greatest influence on each recommended item.

3. Similarity By Category

Someone with similar sports interests may not have the same preference in politics. Even if you and another person are very similar, there will likely be differences in some areas. Socialfeed’s third new indicator shows how similar you are to others, category by category. Socialfeed has 9 categories. They are Business, Tech & Science, Entertainment, Lifestyle, Politics, News, Transportation, Sports and Other. Every feed is assigned to one of these categories. When Socialfeed finds people who are similar to you and calculates the degree of similarity, it compares your activities for each category. Socialfeed then calculates the degree of similarity with each person by category.

SocialFeed is Back!

After a tough period of complete shutdown, the SocialFeed service is back up at http://socialfeed.cc. Thank you for checking the Website and the blog even while we were down.

We are sorry to say that we were unable to recover all of the data. Before telling you what we have recovered and what we have lost, let me explain briefly what happened.

We had two major issues that struck us at the same time: a MySQL database corruption and a physical crash of our server drive.

The physical crash of our server drive was the main reason for the interruption of service. However, shortly before our drive crashed, we discovered that our database was not correctly backing up. The backup shell was dumping only 10-20% of the data. While looking into this issue, we were confronted with the disc crash and were forced to stop everything. What horrible timing!

We sent our HDD to a data recovery service immediately, and fortunately, we were able to recover the data successfully. However, even after the HDD recovery, we found that the database was partially corrupted. We have been working hard to recover lost data from our latest but incomplete backups in order to re-launch our service again with as little data loss as possible.

What we’ve successfully recovered:
- User data
All user data and configurations before the crash were recovered with no loss.
- User relationship data
All the explicit relationships (called “friends” in SocialFeed) were successfully recovered and restored. Implicit friends (called “similar users”) have also been restored.
- Feeds and subscriptions
All the feeds subscribed to by users have been recovered and restored.

What we’ve lost:
- All past stories (approximately 500,000 items).
Stories published before the crash and during the period when the service was down have been lost.

- History
As we have lost all of the past stories, we have also lost all the history related to them.
- Saved and deleted stories
Saved stories and the stories forwarded from your friends are also gone.

We are terribly sorry for these losses and the extended downtime, but we are happy that all user data has been successfully recovered.

SocialFeed has begun collecting newly published stories, and we will release several new features which we developed during the long “vacation”. We hope you will come back again and take advantage of our unique recommendation service. We are working hard to improve the stability of our systems and to make the service more useful and more fun to use!

Humbly yours,

The SocialFeed Team

SocialFeed is Currently Down

Due to a hardware issue, SocialFeed is currently not available. We apologize deeply for the inconvenience. We are working around the clock to fix the problem. While we are uncertain how long SocialFeed will be unavailable, we hope to be back up soon! Thanks for your patience. Please check back in a few days.

Open Beta is up and running, but we’re still fixing some things.

Long time no see. Several days ago, on June 11th, Socialfeed opened as an Open Beta. Socialfeed now has a sign-up feature, and anyone can join.

On June 11th and 12th, we were thrilled to have several hundred first-time users. We were happy, but like all startups, we had some performance problems. Deep apologies to all the newcomers.

I worked hard to fix the problems, and now the site is much better. We made a list of what we can do and will get to these soon.

We’ll be doing the following as soon as possible:

• Find and improve unoptimized code
• Use the RubyOnRails cache feature for the hottest pages
• Optimize SQL

In July, we hope to:

• Replace our server with a much faster one.
• Dedicate separate servers to different roles. Right now, everything is in one place. We’ll get multiple servers and give each one a separate function. This should keep the backend processes from affecting the user experience.
• Optimize DB and SQL
• Load-balancing

We also found several trivial and a few serious bugs that sometimes return an “application error” message. We will work hard to find and fix the bugs, as well as improve the features themselves.

We will also launch a simple feedback form in a few days, and we’re looking forward to hearing from you. Of course, feel free to also post comments to this blog. I will be sure to respond.

Keep connected to your neighbors, and then find some more!

Socialfeed logo

Socialfeed team determined the logo of socialfeed
This is the main logo which is used in pages.

This is the big logo used on the top page.

How To Improve Your Recommendations

SocialFeed recommends news and blogs for you from RSS feeds. The recommendations are made from feeds that your neighbors find interesting. Recommendations are prioritized according to how important the SocialFeed system thinks they are to you. A recommendation’s priority is also based on how many of your neighbors are interested in the same story and how similar these neighbors are to you. In the SocialFeed system these neighbors are called recommendation contributors. If you want to improve the recommendations, you need to find better contributors. There are two ways to do this:

The first way is to let SocialFeed find better or more similar neighbors for you. In order to find better neighbors, SocialFeed must more precisely understand your preferences. You can improve your preferences and profile in the following ways:
1. In the Search & Manage Sources menu, select categories you like using the Directory & Search tab, subscribe to sources you want to read using the Manage Subscription tab, and register feeds using the Add RSS tab.
2. In the Settings menu, select languages and locations.
3. Read whatever you like. Save or post a comment. Forward articles you want to share with your friends. SocialFeed tracks all these activities and uses them to find users similar to you.

The second way to improve your recommendations is to use feedback. This allows you to control your recommendations directly. When you see a story recommended to you under the Recommended tab, you will also see an “I Don’t Like This” button

By clicking this button, the level of similarity between you and the neighbor who recommend the story is decreased. The next time SocialFeed gives you a new recommendation, contributions from this user are given less importance.

These are the two ways to receive better recommendations in SocialFeed.

SocialFeed “Closed beta” started

What is SocialFeed?
The SocialFeed team launched the RSS-based news recommendation service SocialFeed at http://socialfeed.cc on December 11, 2006 as a closed beta. SocialFeed is an experiment in personalized news recommendation services. Our recommendations are personalized and based upon user activities and social elements among users.

Background
WEB 2.0 has led to a lot of social bookmarking and other “democratic” services. But to me, these services are too general, making it difficult to find what I really like. On the other hand, RSS readers provide access to numerous feeds from across the Blogosphere, and everyday so many new posts are created that users can never consume all of them.
So there is a dilemma: The wisdom of the group is often helpful, but sometimes it is too different from our personal preferences. At the same time, individuals are too busy to check all the sources that they may want to read. SocialFeed solves this dilemma through a service that is midway between “democratic” and personalized.

The SocialFeed Solution
I am sure most people have experienced hearing important news by word of mouth from friends, family, or coworkers. Word of mouth is a very accurate source of news because each of your news sources knows you and your preferences, interests and timings. However, word of mouth is a voluntary behavior dependent on people.

SocialFeed gathers your preferences and interests and continuously watches your activities, then automatically finds people with interests similar to yours and measures how similar they are. Based upon this implicit network, SocialFeed recommends news or blogs selected for you.

SocialFeed is personalized and computer-simulated word-of-mouth information sharing. Our goal is to help you discover news and news sources you like to read.

This blog is a SocialFeed project official team blog. My name is Yuki Ishikawa. I am the developer of SocialFeed. On this blog, we plan to post things like…

• Official announcements and updates
• Explanations and instructions for SocialFeed functionality
• Discussion and feedback from users
• Typical blog content