This is a blog examining the use of signaling in applications that provide recommendations as a use case for graph databases.
After reading books like “Freakonomics,” written by pop economic superstars Stephen Dubner and Steven Levitt, and “Outliers,” written by acclaimed New Yorker magazine columnist Malcolm Gladwell, you realize that economic theory can help you see the world from a different perspective. New insights using economic principles are helping to address questions that have previously seemed unanswerable. In his book, “Who Gets What and Why: The New Economics of Matchmaking and Market Design,” Alvin Roth designs successful matching markets where desperate organ recipients find compatible organ donors or needy job seekers find companies looking for candidates with their skillset. Roth is opening up a new way of solving matching problems.
Roth discusses the need for natural signals of interest in successful market designs. This communication, which Roth calls “signaling,” is used to qualify potential matches and determine if the matches are desired (177). For example, bright displays of color are not an evolutionary advantage to avoiding predators for a male peacock, but his plumage certainly allows potential mates to identify him as strong and healthy. Similarly, a signal can be costly for the sender, but it can also help to qualify a match or determine the strength of a match. The value of signaling is seen clearly through use of online dating sites to meet people.
Now that more couples are talking about meeting online, the stigma of using online dating is fading and the industry is growing rapidly. IBISWorld data research2 has shown the industry is showing a steady annual growth rate of 5% between 2010 and 2015 and is projected to continue. At the same time, as the industry grows, the process of finding matches on an internet dating site is growing more complex.
Roth recognizes this phenomenon in his book when he makes a distinction between desirability and interest. A peacock may have a beautiful plume, but it doesn’t describe his interest in a particular mate. Roth writes, “As we’ve seen, in a congested market—one in which it’s impossible to explore every opportunity—it helps to be able to signal not only how desirable you are but also how interested.” (177). In a mating market, interest can be determined and assessed across many categories including job, family, character, religion, hobbies, athleticism, etc. This is why signaling can be quite complex because it can both represent degrees of desirability, but also differing levels of interest.
To design a successful mating market, capturing complex signaling is clearly a requirement. I met my wife on eHarmony™ and one of the things that my wife liked best about me was my clear interest in her. Of course, this was not the end of the story, but it definitely got my foot in the door. This phenomenon is examined more clearly in Behrendt and Tuccillo’s book, “He’s Just Not That Into You,” and the movie of the same name, which my wife referenced many times since we met as a way of filtering out guys based on their lack of quality signaling. Signaling early and confidently communicates a strong message.
Tinder™ is an online application that uses signaling to indicate matches between users. Just like the peacocks, you indicate interest in someone simply by swiping after looking at their picture. A swipe on a picture signals interest, and if both sides swipe right, a match is created. More recently, Tinder realized that qualifying based on a single factor was not sufficient, so now it allows more data in user profiles, which enables a smarter algorithm for recommending potential matches. Regarding the update, the company has measured “a ‘meaningful increase’ in matches and the quality of conversations.”4 This shows that more information when signaling leads to a more successful system.
Providing a way to add signals between users and recommending quality matches by examining those signals and other related factors is the specialty of a graph database. By using a graph database like InfiniteGraph, you can store information for complex signal types and use these types to recommend and discover the matches. You will find that this type of graph data is a perfect example of what makes graph databases necessary.
Figure 2: A Type of a Signal Edge
public class Attraction extends Signal
{
private int level;
public Attraction (int level)
{
this.level = level;
}
public int getLevel()
{
fetch();
return level;
}
}
Recommendations based on multiple degrees of separation, typical in exchange markets, are a common use case for graph databases, because it avoids expensive and complex JOIN operations on top of a traditional relational database. JOIN operations are used to perform linking between tables based on a common foreign key and are necessary to perform multi-hop navigations. JOINs are extremely computationally expensive.
With InfiniteGraph, you don’t have to deal with the complexity and performance bottlenecks that JOINs add. Instead, you can focus on the data and algorithms based on the data that add value to your application. Check out my exchange blog series for more information about the usefulness of graph databases in designing a market. If you are interested in evaluating our graph database, please check out our products page and contact us.