Web Reputation Systems

During my research into user reputation systems, I came across the book Building Web Reputation Systems by Randy Farmer and Bryce Glass. In the book's preface, the authors answer the question, why write a book about reputation?

"We wrote this book because we saw how critical reputation has become to the survival and growth of the Web. Though there are many academic research papers on specific generational algorithms and social effects of reputation systems, we couldn’t find a single book that put it all in context - describing it as a separate domain of knowledge, complete with a grammar, emerging best-use patterns, and recurring antipatterns."

Thankfully, the authors carried out their vision to produce this tome. It has been an inspiration in its role as a starting point for someone who has yet to develop any kind of advanced user reputation features for an online application.

We all know the big sites like Google, Facebook, Amazon, Yahoo, eBay, etc. keep monolithic databases of our behavior and crunch them with algorithms created by armies of PhDs to add reputation-based features to their services. The question is, how can smaller teams, or even single-dev operations, build reputation features into their social discovery, online dating and community software? 

That's the question I had in my mind when searching for help with one of the "alpha stage" projects I work on in my spare time. 

SocialD8.com is a experimental project I've created as an outlet for my ideas regarding an "interest-based singles-based community management system" (IBSCMS). A goal for this dating site testbed is the continuous improvement of the user experience by weeding out members who aren't serious about finding someone with shared interests. By employing a user reputation strategy, I hope to increase the number, and effectiveness, of "trust cues" that members can use to make value judgments about other members.

As reputation systems can vary from simple to advanced in their scope and capabilities, it can be a challenge to grasp everything at once. It is for this reason that I liked a quote from another book that I am using for my research, Designing Social Interfaces: "The trick is to have any kind of framework to start from, not to have a definitive and comprehensive model."

What is meant by "reputation"?

A user's reputation in an online community is defined this way by the book's authors: "Information used to make a value judgment about an object or person."

Here's an expanded definition from Wikipedia: "Reputation of a social entity (a person, a group of people, an organization) is an opinion about that entity, typically a result of social evaluation on a set of criteria. It is important in education, business, and online communities. Reputation may be considered as a component of identity as defined by others."

A reputation system is an important tool that helps dating site users make decisions about whether or not to engage with another single. If users are more informed about the intent of other users through reputation scores, they are more likely to engage with the other members.

The book's authors stress that reputation always "takes place within context". Within an online dating site, the context is "trust".  Members want to trust that the person on the other side is real, not trying to spam or scam them. On free dating sites, this is significantly more important, especially for abuse mitigation. When a member has engaged in the effort to build reputation points, it signals to others an intent to be serious and to not waste time.

Sample Reputation-based Features

Often, it is best to learn through example. So what are some examples of the kind of features we'd expect in a reputation-aware application? I'll only mention them briefly... going into details about each one is beyond the scope of this article.

• The first one is StackOverflow's Badge System, which can be found here. Here's a quick description of how the Stack Overflow reputation system works.
• The eBay Seller Feedback metric is a widely known rating that people have come to trust.
• Amazon 's review system is a powerful reputation system. The products have a reputation and the reviewers have a reputation. This dual-trust system is one I use quite often when deciding on products to buy.
• Take a look at Yelp's badge list.

Here's a list of reputation patterns listed in the Building Web Reputation Systems book:

• Vote to promote
• Content rating and ranking
• Content comments and reviews
• Leaderboards
• Incentive karma
• Quality karma
• Competitive karma
• Abuse scoring

Yahoo! has made available for developers the Yahoo! Reputation Patterns library:

• The Competitive Spectrum
• Named Levels
• Numbered Levels
• Identifying Labels
• Points
• Collectible Achievements
• Ranking

Conclusion

At this stage, I am still in the discovery phase of evaluating what behaviors to include in the SocialD8 reputation system and the weightings to be assigned to each measured unit of user engagement. Reputation metrics are clearly going to play an increasingly important role in social discovery sites, especially in the online dating space.

The more I dive into this topic, the more exciting the possibilities become as I realize we're at the nascent stage of embedding advanced reputation features into online applications.

- - - - - -

More from the Wikipedia definition: "A reputation system computes and publishes reputation scores for a set of objects (e.g. service providers, services, goods or entities) within a community or domain, based on a collection of opinions that other entities hold about the objects. The opinions are typically passed as ratings to a reputation center which uses a specific reputation algorithm to dynamically compute the reputation scores based on the received ratings."

Randy Farmer's Google Tech Talks presentation on YouTube: Building Web Reputation Systems.

Recent Posts...