Amazon Scholar wins “highest honor” in knowledge discovery and data mining

Thorsten Joachims answers 3 questions about the work that earned him the award.

This week, at the Association for Computing Machinery’s annual conference on Knowledge Discovery and Data Mining (KDD), Thorsten Joachims, a Cornell University professor of computer science and an Amazon Scholar, received the conference’s Innovation Award. SIGKDD, the ACM special-interest group that organizes the conference, describes the award as “the highest honor for technical excellence in the field of knowledge discovery and data mining”.

Thorsten Joachims
Thorsten Joachims, a Cornell University professor of computer science and an Amazon Scholar.

The award citation acknowledges Joachims’s “influential work studying human biases in information retrieval, SVM [support vector machines], and structured output prediction” and particularly his “methods for eliciting reliable preferences from implicit feedback, methods for unbiased learning-to-rank, and ranking methods that provide fairness guarantees.”

Amazon Science asked Joachims three questions about the work that earned the award — particularly as it pertains to his research at Amazon.

Q: What is the problem of “human biases in information retrieval”?

Much of my work has been about trying to learn from human behavior, particularly in systems that provide rankings or that provide recommendations. A great source of feedback that these systems provide is whether someone clicks on a result or reformulates a query and eventually consumes something. That provides a lot of data, and unlike more traditional ways of training these system, it’s not what some expert thinks is relevant to this query. It actually reflects what the users think — what’s the right answer to that query or what’s helpful to them.

The problem with learning from this implicit feedback is that the system biases how people behave. Things that are ranked highly will get a lot more exposure than things that are way down below, and that also affects what people can click or what people will purchase. So by the actions that the system takes, it is also contaminating the data. Something in position one gets the most clicks, so it stays in position one. It becomes this self-reinforcing cycle, which means that something that’s pretty bad can stay in position one while something that’s pretty good never gets discovered.

The question is, how do you deal with the biases that the systems introduce? One general insight is to view these systems as agents that interact with people: they’re not just collecting data; they’re taking an action, like making a recommendation. And what we observe is how people react to that intervention.

This means that system is a lot like a controlled randomized trial in medicine. You give a treatment to a patient, and you see how the patient reacts to that treatment, but you don’t get to see what would have happened if you had given that patient another treatment.

The same with a recommender system or ranker system: you see what happens if you make a particular intervention — recommend this movie, and the person watches it or not — but you don’t get to see what would have happened if you had recommended a different item. What we’ve brought to recommender systems is this idea that you want to treat them in the same way, from a statistical perspective, as a controlled randomized trial.

Now, in some ways this problem is easier than in medicine. We get a lot more data, and there’s a lot less risk. But in some sense it’s also harder. In medicine, you might have three different treatments, whereas in recommender systems, every item in your database is a potential treatment. We have millions of them, so dealing with complexity and the scale of the problem is challenging.

Something in position one gets the most clicks, so it stays in position one. It becomes this self-reinforcing cycle.

There are two ways you can approach this problem: the online way and the offline way. In the online way, you constantly try out new interventions, see how people react, and then tweak your policy step by step, always interactively running these experiments. That’s called online learning and, in particular, contextual-bandit online learning

In a sense, online learning is wasteful, and it potentially has a negative impact on the customer, in that you’re potentially trying out things multiple times that are not actually that good.

But we already have terabytes of existing data where we know we’ve taken that action in this context for that customer, and the customer was happy. Can we recycle all of this old data and use that for machine learning, instead of trying things out over and over again?

One of the things we’ve developed are these batch learning methods, which you can think about as learning from a controlled randomized trial in hindsight. Once you have the data, ask the question, “What would have been the best policy if I were able to rewind time and go back to when the data was collected?” I think these offline algorithms are particularly promising.

Q: Do you use the same approach in the “learning-to-rank” setting mentioned in the award citation?

Learning to rank is one particular type of feedback. Contextual bandit is more like, you ask Alexa to play music, and Alexa has to play something for you. It picks exactly one action: it plays one track, and the user liked it or didn’t like it. The ranking setting is a little more forgiving. You present a ranking of items, so even if you don’t nail the top one, you can still get feedback if the user is patient and goes down the ranking.

But if the user doesn’t click on something, that could have two causes. One cause is the user didn’t like it. The other cause is that the user just didn’t see it; the user didn’t go far enough down to discover that item.

So the additional complication, compared to the contextual-bandit setting, is that you have to tease apart this ambiguity. We’ve come up with techniques where you can at least in expectation tease these two causes apart. Even though you can’t do it for any individual impression, you can say, “In expectation, I know that lack of seeing an item is responsible for that many missing clicks, and lack of relevance is responsible for the rest.” So similar techniques that come from controlled randomized trials can also be brought to bear on this problem.

Q: What is “structured output prediction”, which the citation also mentions?

Many machine learning problems are formulated as binary classifications — predicting yes or no — or as regression problems, where you just predict a number — 5.7 or something.

But for many other problems, you’re predicting a structured object. Rankings are an example of a structured object, where the thing that you’re predicting is a combinatorial object. It’s a permutation.

You want to model dependencies in this ranking. For example, if you have the query “Michael Jordan”, that’s ambiguous. It could mean the basketball player; it could mean the statistician; it could mean the actor.

Maybe the basketball player is the most likely interpretation, but filling your top 10 results only with links about the basketball player is probably not the right thing to do, because not everybody was looking for that.

You want to model dependencies: if I put the first links about the basketball player, what’s the best thing to put next? Maybe the next most popular intent is the actor. You want to make this prediction of what you put into the ranking as a prediction of all these items that are dependent on each other.

That gives you these machine learning problems where the thing that you’re predicting is one element of this huge combinatorial space of all possible permutations over documents, which are more than there are atoms in the universe. And you still want to learn these models, and you want to efficiently compute what’s the best ranking to present.

That’s a problem that’s relevant to Amazon. It’s also relevant to a lot of other problems, like predicting the structure of a protein. You have the sequence, and you want to predict how it folds. You really have to model all the dependencies, how things interact in the protein.

Or it’s relevant to natural-language processing — predicting the constituents of the semantic parse of a sentence, for example. You need to take into account how all of the constituents of the sentence relate to each other. So it’s really relevant to a lot of prediction problems.

Related content

US, WA, Seattle
Amazon internships are full-time (40 hours/week) for 12 consecutive weeks with start dates in May - July 2023. Our internship program provides hands-on learning and building experiences for students who are interested in a career in hardware engineering. This role will be based in Seattle, and candidates must be willing to work in-person. Corporate Projects (CPT) is a team that sits within the broader Corporate Development organization at Amazon. We seek to bring net-new, strategic projects to life by working together with customers and evolving projects from ZERO-to-ONE. To do so, we deploy our resources towards proofs-of-concept (POCs) and pilot programs and develop them from high-level ideas (the ZERO) to tangible short-term results that provide validating signal and a path to scale (the ONE). We work with our customers to develop and create net-new opportunities by relentlessly scouring all of Amazon and finding new and innovative ways to strengthen and/or accelerate the Amazon Flywheel. CPT seeks an Applied Science intern to work with a diverse, cross-functional team to build new, innovative customer experiences. Within CPT, you will apply both traditional and novel scientific approaches to solve and scale problems and solutions. We are a team where science meets application. A successful candidate will be a self-starter comfortable with ambiguity, strong attention to detail, and the ability to work in a fast-paced, ever-changing environment. As an Applied Science Intern, you will own the design and development of end-to-end systems. You’ll have the opportunity to create technical roadmaps, and drive production level projects that will support Amazon Science. You will work closely with Amazon scientists, and other science interns to develop solutions and deploy them into production. The ideal scientist must have the ability to work with diverse groups of people and cross-functional teams to solve complex business problems.
US, WA, Seattle
Amazon internships are full-time (40 hours/week) for 12 consecutive weeks with start dates in May - July 2023. Our internship program provides hands-on learning and building experiences for students who are interested in a career in hardware engineering. This role will be based in Seattle, and candidates must be willing to work in-person. Corporate Projects (CPT) is a team that sits within the broader Corporate Development organization at Amazon. We seek to bring net-new, strategic projects to life by working together with customers and evolving projects from ZERO-to-ONE. To do so, we deploy our resources towards proofs-of-concept (POCs) and pilot programs and develop them from high-level ideas (the ZERO) to tangible short-term results that provide validating signal and a path to scale (the ONE). We work with our customers to develop and create net-new opportunities by relentlessly scouring all of Amazon and finding new and innovative ways to strengthen and/or accelerate the Amazon Flywheel. CPT seeks an Applied Science intern to work with a diverse, cross-functional team to build new, innovative customer experiences. Within CPT, you will apply both traditional and novel scientific approaches to solve and scale problems and solutions. We are a team where science meets application. A successful candidate will be a self-starter comfortable with ambiguity, strong attention to detail, and the ability to work in a fast-paced, ever-changing environment. As an Applied Science Intern, you will own the design and development of end-to-end systems. You’ll have the opportunity to create technical roadmaps, and drive production level projects that will support Amazon Science. You will work closely with Amazon scientists, and other science interns to develop solutions and deploy them into production. The ideal scientist must have the ability to work with diverse groups of people and cross-functional teams to solve complex business problems.
US, MA, Westborough
Are you inspired by invention? Is problem solving through teamwork in your DNA? Do you like the idea of seeing how your work impacts the bigger picture? Answer yes to any of these and you’ll fit right in here at Amazon Robotics. We are a smart team of doers that work passionately to apply cutting edge advances in robotics and software to solve real-world challenges that will transform our customers’ experiences in ways we can’t even imagine yet. We invent new improvements every day. We are Amazon Robotics and we will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun. Amazon Robotics is seeking interns and co-ops with a passion for robotic research to work on cutting edge algorithms for robotics. Our team works on challenging and high-impact projects, including allocating resources to complete a million orders a day, coordinating the motion of thousands of robots, autonomous navigation in warehouses, identifying objects and damage, and learning how to grasp all the products Amazon sells. We are seeking internship candidates with backgrounds in computer vision, machine learning, resource allocation, discrete optimization, search, and planning/scheduling. You will be challenged intellectually and have a good time while you are at it! Key job responsibilities • Identifying creative solutions for challenging research problems in robotics and computer vision • Developing software solutions to test hypotheses and demonstrate new functionality • Prototyping concepts to collect data and measure performance • Writing code and unit tests and integrating code with other software and hardware components • Utilizing Amazon Robotics and Amazon engineering tools, processes and technologies • Delivering a final presentation to managers and engineers on the successes and challenges of their internship and the business value they have contributed
US, MA, Westborough
Are you inspired by invention? Is problem solving through teamwork in your DNA? Do you like the idea of seeing how your work impacts the bigger picture? Answer yes to any of these and you’ll fit right in here at Amazon Robotics. We are a smart team of doers that work passionately to apply cutting edge advances in robotics and software to solve real-world challenges that will transform our customers’ experiences in ways we can’t even imagine yet. We invent new improvements every day. We are Amazon Robotics and we will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun. Amazon Robotics is seeking interns and co-ops with a passion for robotic research to work on cutting edge algorithms for robotics. Our team works on challenging and high-impact projects, including allocating resources to complete a million orders a day, coordinating the motion of thousands of robots, autonomous navigation in warehouses, identifying objects and damage, and learning how to grasp all the products Amazon sells. We are seeking internship candidates with backgrounds in computer vision, machine learning, resource allocation, discrete optimization, search, and planning/scheduling. You will be challenged intellectually and have a good time while you are at it! Please note that by applying to this role you would be considered for Applied Scientist summer intern, spring co-op, and fall co-op roles on various Amazon Robotics teams. These teams work on robotics research within areas such as computer vision, machine learning, robotic manipulation, navigation, path planning, perception, artificial intelligence, human-robot interaction, optimization and more.
US, CA, Palo Alto
The Amazon Search team creates powerful, customer-focused search solutions and technologies. Whenever a customer visits an Amazon site worldwide and types in a query or browses through product categories, Amazon Search services go to work. We design, develop, and deploy high performance, fault-tolerant distributed search systems used by millions of Amazon customers every day. We’re seeking a Principal Scientist with a deep expertise in Search Science. Your responsibilities will include everything from developing and prototyping innovative machine learning, and deep learning algorithms to implementing, testing, and supporting full solutions in a production environment. We are looking for innovators who can contribute to advancing search technology on what’s scientifically possible while remaining committed to creating world-class products. Joining this team, you’ll experience the benefits of working in a dynamic, entrepreneurial environment, while leveraging the resources of Amazon.com (AMZN), Earth's most customer-centric company one of the world's leading internet companies. We provide a highly customer-centric, team-oriented environment in our offices located in Palo Alto, California. Key job responsibilities As a hands-on leader of this team, you’ll be responsible for defining key research questions, identifying relevant data, adopting or proposing innovative machine learning solutions conducting rigorous experiments, publishing results and working with the engineering team to deploy these solutions. As a strategic leader, you will identify investment opportunities, develop long term strategies, and propose, prioritize and deliver on goals. You’ll also participate in organizational planning, hiring, mentorship and leadership development. You will be technically fearless and with a passion for building scalable science and engineering solutions. You will serve as a key scientific resource in full-cycle development (conception, design, implementation, testing to documentation, delivery, and maintenance). About the team Starting in 2009, the Visual Search & Augmented Reality team has thus far launched many visual search solutions on the Amazon App that use computer vision and machine learning/deep learning to help customers complete their shopping missions more easily; multiple internal teams at Amazon (devices, Kindle, Seller services, etc.) also use our libraries and APIs to deliver solutions to their own customers. We are a full stack shop, and our team capabilities cover the whole solution spectrum, ranging across applied science, large scale engineering services, product management, UX design, and mobile app development for iOS and Android.
US, MN, Minneapolis
AWS Central Economics is an interdisciplinary team on the cutting edge of economics, statistical analysis, and machine learning whose mission is to solve problems that have high risk with abnormally high returns. Our team leverages the strengths of our scientists to build solutions for some of the toughest business problems here at Amazon AWS. We are looking for an exceptionally talented, seasoned, and motivated Economist to manage a team of economists and data scientists to drive the science for AWS. Key job responsibilities Manage a team of economists and data scientists to deliver actionable economic analyses to business leaders, provide leadership on the economics and science used in the analyses, and engage with business leaders to identify challenges AWS faces that call for in-depth economic analyses and to ensure the analyses have their intended impact.
LU, Luxembourg
&ltHire Relocation Requisition - not for posting> Provides insights to leadership on improving Supply Chain cost and Speed by using Data Science and Analytics techniques. Build Dashboards and models to industrialize these findings at scale.
US, VA, Arlington
The People eXperience and Technology Central Science Team (PXTCS) uses economics, behavioral science, statistics, and machine learning to proactively identify mechanisms and process improvements which simultaneously improve Amazon and the lives, wellbeing, and the value of work to Amazonians. We are an interdisciplinary team that combines the talents of science and engineering to develop and deliver solutions that measurably achieve this goal. We are looking for economists who are able to work with business partners to hone complex problems into specific, scientific questions, and test those questions to generate insights. The ideal candidate will work with engineers and computer scientists to estimate models and algorithms on large scale data, design pilots and measure their impact, and transform successful prototypes into improved policies and programs at scale. We are looking for creative thinkers who can combine a strong technical economic toolbox with a desire to learn from other disciplines, and who know how to execute and deliver on big ideas as part of an interdisciplinary technical team. Ideal candidates will work closely with business partners to develop science that solves the most important business challenges. They will work in a team setting with individuals from diverse disciplines and backgrounds. They will serve as an ambassador for science and a scientific resource for business teams, so that scientific processes permeate throughout the HR organization to the benefit of Amazonians and Amazon. Ideal candidates will own the data analysis, modeling, and experimentation that is necessary for estimating and validating models. They will work closely with engineering teams to develop scalable data resources to support rapid insights, and take successful models and findings into production as new products and services. They will be customer-centric and will communicate scientific approaches and findings to business leaders, listening to and incorporate their feedback, and delivering successful scientific solutions. Key job responsibilities Use causal inference methods to evaluate the impact of policies on employee outcomes. Examine how external labor market and economic conditions impact Amazon's ability to hire and retain talent. Use scientifically rigorous methods to develop and recommend career paths for employees. A day in the life Work with teammates to apply economic methods to business problems. This might include identifying the appropriate research questions, writing code to implement a DID analysis or estimate a structural model, or writing and presenting a document with findings to business leaders. Our economists also collaborate with partner teams throughout the process, from understanding their challenges, to developing a research agenda that will address those challenges, to help them implement solutions. About the team We are a multidisciplinary team that combines the talents of science and engineering to develop innovative solutions to make Amazon Earth's Best Employer.
US, WA, Seattle
The People eXperience and Technology Central Science Team (PXTCS) uses economics, behavioral science, statistics, and machine learning to proactively identify mechanisms and process improvements which simultaneously improve Amazon and the lives, wellbeing, and the value of work to Amazonians. We are an interdisciplinary team that combines the talents of science and engineering to develop and deliver solutions that measurably achieve this goal. We are looking for economists who are able to apply economic methods to address business problems. The ideal candidate will work with engineers and computer scientists to estimate models and algorithms on large scale data, design pilots and measure their impact, and transform successful prototypes into improved policies and programs at scale. We are looking for creative thinkers who can combine a strong technical economic toolbox with a desire to learn from other disciplines, and who know how to execute and deliver on big ideas as part of an interdisciplinary technical team. Ideal candidates will work in a team setting with individuals from diverse disciplines and backgrounds. They will work with teammates to develop scientific models and conduct the data analysis, modeling, and experimentation that is necessary for estimating and validating models. They will work closely with engineering teams to develop scalable data resources to support rapid insights, and take successful models and findings into production as new products and services. They will be customer-centric and will communicate scientific approaches and findings to business leaders, listening to and incorporate their feedback, and delivering successful scientific solutions. Key job responsibilities Use causal inference methods to evaluate the impact of policies on employee outcomes. Examine how external labor market and economic conditions impact Amazon's ability to hire and retain talent. Use scientifically rigorous methods to develop and recommend career paths for employees. A day in the life Work with teammates to apply economic methods to business problems. This might include identifying the appropriate research questions, writing code to implement a DID analysis or estimate a structural model, or writing and presenting a document with findings to business leaders. Our economists also collaborate with partner teams throughout the process, from understanding their challenges, to developing a research agenda that will address those challenges, to help them implement solutions. About the team We are a multidisciplinary team that combines the talents of science and engineering to develop innovative solutions to make Amazon Earth's Best Employer.
US, WA, Seattle
Amazon is looking for talented Postdoctoral Scientists to join our global Science teams for a one-year, full-time research position. Postdoctoral Scientists will innovate as members of Amazon’s key global Science teams, including: AWS, Alexa AI, Alexa Shopping, Amazon Style, CoreAI, Last Mile, and Supply Chain Optimization Technologies. Postdoctoral Scientists will join one of may central, global science teams focused on solving research-intense business problems by leveraging Machine Learning, Econometrics, Statistics, and Data Science. Postdoctoral Scientists will work at the intersection of ML and systems to solve practical data driven optimization problems at Amazon scale. Postdocs will raise the scientific bar across Amazon by diving deep into exploratory areas of research to enhance the customer experience and improve efficiencies. Please note: This posting is one of several Amazon Postdoctoral Scientist postings. Please only apply to a maximum of 2 Amazon Postdoctoral Scientist postings that are relevant to your technical field and subject matter expertise. Key job responsibilities * Work closely with a senior science advisor, collaborate with other scientists and engineers, and be part of Amazon’s vibrant and diverse global science community. * Publish your innovation in top-tier academic venues and hone your presentation skills. * Be inspired by challenges and opportunities to invent cutting-edge techniques in your area(s) of expertise.