Strands Blog

Presented by Dr. Francisco Martin, Founder and CEO of Strands, the first keynote at the 3rd annual ACM Conference on Recommender Systems was delivered this morning and has been generating some buzz on Twitter.

The number of online services providing users with real-time recommendations has increased exponentially over the last few years. Recommender Systems that were originally only accessible to a limited number of high-tech companies are now widely available through a growing number of both technical choices and vendors. The acceptance, however, of automatically delivered recommendations by users depends on numerous factors that go far beyond the algorithms that constitute the major focus of researchers.

Over the past 10 years, Strands has been creating and operating recommender systems in a multitude of domains, ranging from digital music to fitness plans and personal finance management, and in a multitude of business settings ranging from lightweight integrations to highly-coupled integrations within secure bank environments.

As summarized by Neal Lathia of MobBlog, below are the top 10 lessons we’ve learned developing, deploying, and operating real-world recommender systems:

• Lesson 1. Make sure a recommender is really needed! Do you have lots of recommendable items? Many diverse customers?… also think Return-on-Invesment… a more sophisticated recommender may not deliver a better ROI.
• Lesson 2. Make sure the recommendations make strategic sense. Is the best recommendation for the customer also the best for the business? What is the difference between a good and useful recommendation? Good recommendations vs useful recs; obvious recommendations may not be useful; risky recs may deliver better long-term value.
• Lesson 3. Choose the right partner! Select the right rec vendor vs hire some #recsys09 students. If you are a big company the best thing you can do is organize a contest.
• Lesson 4. Forget about cold-start problems (!) …. just be creative. The internet has all the data that you need (somewhere…).
• Lesson 5. Get the right balance between data and algorithms. 70% of the success of a recommendation system is in the data, the other 30% in the algorithm.
• Lesson 6. Finding correlated items is easy but deciding what, how, and when to present them to the user is hard… or don’t just recommend for the sake of it. Remember, user attention is a scarce and valuable resource. Use it wisely! … dont make a recommendations to a customer who is just about to pay for items at the checkout! User interface should get at least 50% of your attention.
• Lesson 7 Don’t waste time computing nearest neighbours (use social connections)… just mine the social graph. Might miss useful connections?
• Lesson 8 Don’t wait to scale!
• Lesson 9: Choose the right feedback mechanism. Stars vs thumbs …. the YouTube problem. More research on implicit and other feedback mechanisms is needed. The perfect rating system is no rating system! … focus on the interface.
• Lesson 10 Measure Everything! … business control and analytics is a big opportunity here.

Keynote Takeaway

Think about application context; Focus on interface as much as algs; be creative with startup data. … the UI needs to get the lion’s share of the effort (50%) compared to algorithms (5%), knowledge (20%), analytics (25%).

Strands.com will host the inaugural edition of the Strands 5k in Corvallis on October 24th. The race will feature a community 5k, a children’s race, post-race party and a prize purse totaling $13,000. Supporting sponsors of the race include Emery & Sons Construction, Eufora, and Marrakesh Salon.

All race proceeds benefit Childhelp, an organization dedicated to helping victims of child abuse and neglect. Childhelp’s mission is to meet the physical, emotional, and spiritual needs of abused, neglected and at-risk children through advocacy, prevention, treatment, and community outreach. Find more ways to get involved or information on how to donate directly through the Childhelp.org website.

Prize money is only available to U.S. Citizens and will be distributed as follows for the men’s and women’s open field:

• 1st place–$3,000
• 2nd place–$1,500
• 3rd place–$750
• 4th place–$600
• 5th place–$400
• 6th place–$250

The 5k race will begin at 9:00 a.m. PST on October 24. The children’s race will begin at 10:00 a.m., followed by an awards ceremony and post-race party outside Strands headquarters at Central Park in Corvallis.

Online registration is $15.00 and race day registration is $20.00. For more information and a race registration link, visit the race website at Strands5k.com, call the race hotline at (541) 740-5306, or email us at strands5k@strands.com.

Strands Recommender welcomes it’s newest client, CostumesInc.com, in time to offer recommendations for Halloween.

CostumesInc.comquickly deployed recommendations powered by Strands Recommender throughout their site just in time to offer recommendations during their busiest time. The recommendations are offered on the product and shopping cart pages.

Additionally, Costumes Inc. uses Strands recommendations to power each of its tabs for accessories, wigs and hats, makeup and related items shown on individual product pages. Click here to get your Halloween Costume!

We are pleased to invite you to participate in this first workshop on Recommender-based Industrial Applications, to be held on Sunday, October 25th, 2009 in New York City, USA, in conjunction with The 3rd ACM Conference on Recommender Systems (New York City, October 22-25, 2009).

Complete workshop schedule and details can be found here.

Recommender systems help users deal with large amounts of information by assisting them in their decision-making processes.  Recommender systems have been applied very successfully in a variety of domains, and one of their most exciting characteristics is that they combine interest from industry while posing very interesting research challenges, having significant impact on end user experience, and on business.

Motivation and Topics

Recommender-based industrial applications have hit the mainstream: they are being used today by millions of users in a variety of domains (e.g., film, music, book purchases, travel, advertising, etc.), having significant impact on end users and contributing to the generation of millions of dollars of revenue in some industries.

In addition, the explosion of social network websites, on-line user-generated content platforms, and tremendous growth in computational power of mobile devices are generating incredibly large amounts of user data, and an increasing desire of users to find what best matches their personal interests.  This is creating new technical challenges as well as new business opportunities.  However, in spite of significant progress in the research community, there are still important gaps in making accurate and timely recommendations in industrial applications.

This workshop aims at bringing the gap between academic researchers and industry practitioners on the area of Recommender Systems.  We are interested both in research work that faces real industry problems, and in industry cases that create research challenges.

We aim to discuss problems the industry is facing in real-world applications when applying recommendation and personalization techniques.  The goal is to bring together top-level researchers from academia and recommendation practitioners.  Ideally the workshop should serve to establish new collaborations and projects involving academic researchers and companies.

Topics of interest include industrial applications and research challenges related but not limited to:

• case studies of recommender system implementations and deployments
• personalization vs. content-based vs. collaborative-based applications
• evaluation and user studies of recommender systems
• recommendations in the long tail
• user preferences and recommender systems
• social and cultural impact of recommender systems
• security and privacy in recommender systems
• sparsity and lack of data in recommender systems
• context-aware and ubiquitousness in recommender systems
• user interfaces and usability issues in recommender systems
• scalability in large recommender systems
• new business models related to recommendation and personalization
• search vs. recommendation

Workshop Chairs

• Dr. Alejandro Jaimes, Telefonica Research
• Dr. Marc Torrens, Strands

Please address questions to recsys [at] strands.com.

Team Strands athlete Elva Dryer has been selected to the IAAF World Half Marathon Championships team.  The race in Birmingham, U.K. will be the fifteenth time that Elva has represented the United States in international competition.

The two-time Olympian (2000, 2004) has participated in 7 World Cross Country Championships, 3 World Outdoor Track and Field Championships, as well as the World University Games and Goodwill Games.  In four of her seven World Cross appearances, she helped lead Team USA onto the podium, with team bronze medal finishes in 1998 (4km), 2000 (8km), and 2003 (8km), and team silver in Portugal in 2002 (8km).

Dryer, who has a personal best of 1:11:42 from her 2007 USA Half Marathon Championships title, was selected based on her 7th place finish at the USA 20 km Championships earlier this month.

Dryer will team up wth Amy Yoder-Begley (Portland, Oreg.), Desiree Davila (Rochester Hills, Mich.), Serena Burla (Baldwin, Mo.), and Amy Hastings of Mammoth Lakes, California.

“I am honored to be representing the US in the World Half Marathon Championships in Birmingham”, Dryer said. “My goal is to help bring home a top team finish.”

The invitation came just days after Elva elected to forgo racing in the US Marathon Championships in what she says will be an effort to “focus once again on reconnecting with the shorter distances”.

The race will begin at 9:00 a.m. (Birmingham) on October 11, with a web cast from Universal Sports. For more information, visit the official IAAF race website.

Team Strands athlete Jason Hartmann ran away from the field in the last 5 km at the Medtronic Twin Cities Marathon to claim the overall title in a personal best of 2 hours, 12 minutes, 9 seconds.

Hartmann paced himself a few seconds ahead of the lead pack for much of the race, before making a decisive move to the front on Summit Avenue in Saint Paul, dropping the 2004 Twin Cities Marathon Champion Augsutus Mbusya of Kenya (2:13:03), John Njorge (2:13:26), and Joseph Mutinda (2:13:47).  It was the first marathon for Hartmann since his 2:15:27 tenth place finish in the Olympic Marathon Trials.

The win caps a strong road-racing season for Hartmann, which included a runner-up finish in the USA 25 km Championships, ninth place in the USA 10 km Championships, and a sixth place finish in the USA 20 km Championships last month.

For complete results, visit the Medtronic Twin Cities Marathon results page.

Team Strands Update: USA 10-mile Championships

Team Strands athletes Mike Sayenko (48:28), Mike Reneau (48:53), and Josh Glaab (49:20) raced in the US 10-mile Championships, placing 11th, 15th and 18th, respectively.

Complete results for the 10-mile Championships are available here.