Tuesday, June 10

The Rise of Intention and Preference Machines

Yesterday, I mentioned Eric Horvitz's presentation Machine Learning, Reasoning, and Intelligence in Daily Life: Directions and Challenges.

He spends a good deal of his presentation talking about "preference machines" which include recommendation systems. Intention machines are services that use models to predict activities and goals. In short, they uses past history to predict future behavior.

First, an excerpt from the mobile arena, the "Predestination" project that predicts driver destinations.
We have been exploring the uses of the data in learning and reasoning systems, including the construction of a system that can predict and then harness drivers’ likely destinations, given initial driving trajectories [Krumm and Horvitz, 2006]. Beyond geocentric intention machines, we have been exploring the feasibility of building geocentric preference machines, that perform geocentric collaborative filtering: Given sets of sensed destinations of multiple people and the sensed destinations of a particular driver, what places, unvisited previously by that driver, might be of interest, and how and when might the driver be best informed (e.g., by hearing a paid advertisement when he or she is approaching such destinations).
Challenges in Learning and supervision
Priorities research explored a middle ground of allowing users to become more involved with in-stream supervision. In versions of Priorities, users could inspect and modify in-stream supervision policies. Such awareness and potential modification allows the in-stream supervision to become a grounded collaboration between the machine and user...

Challenging areas of research include developing a better understanding of the best approaches to constructing generic models that can provide valuable, usable initial experiences with intelligent applications and services, but that allow for efficient adaptation downstream with a user’s explicit training efforts or in-stream supervision. Research may lead to deeper insights about setting up systems for “ideal adaptability” given expectations about the nature of different kinds of environments, and adaptations, given the users and uses.
Machines and humans need to learn to work together. Sometimes machines can help us make decisions, but one key challenge is to translate the machine's recommendation into a rationale that humans can understand and for this to begin a "dialogue" to correct mistakes and provide more accurate predictions.

This barrier is one reason that Google does not use ML for their core ranking algorithm, see the recent post "Are Machine-Learned Models Prone to Catastrophic Errors?" for an enlightening interview with Google's Peter Norvig. Anand relates,
Peter tells me that their best machine-learned model is now as good as, and sometimes better than, the hand-tuned formula on the results quality metrics that Google uses... Google's search team worries that machine-learned models may be susceptible to catastrophic errors on searches that look very different from the training data. They believe the manually crafted model is less susceptible to such catastrophic errors on unforeseen query types.

Citysense and Context That Matters for Search and Recommendation

The big news is Steve Jobs announcement of the new iPhone 2.0 with 3G and GPS support and, more importantly, a big price drop. This means lots more users for the Apple App Store that will soon be opening for users to download and buy software. There will surely be a deluge of applications that leverage the GPS support, such as Loopt, a mobile social network application that will be available for free.

What does this have to do with search? Everything. Location and temporal context is critical to providing relevant information that I need right now. For example, to find alternative movie theatres when I'm at the mall and the movie I wanted to see was sold out.

There has been a lot of talk about context in Information Retrieval, the IRIX 2004 Workshop, CIR 2007, and the work of Susan Dumais using context, among much research. However, much of this has focused on using your documents, browsing history, and e-mail as context. The prevalence of location-enabled mobile phones is beginning to change all that. The upcoming Workshop on Mobile Information Retrieval (MobIR) at SIGIR looks quite interesting.

A new startup that made me think think differently about mobile context, CitySense by Sense Networks. CitySense (see coverage from O'Reilly and GigaOm) is the latest "social navigation and nightlife discovery application" utilizing real-time and historical location data to show "hotspots" in San Franscisco. What's remarkable is the utilization of not just your location, but also the real-time location of people around you. Sense Networks applies machine learning algorithms to analyze the historical information and identify interesting traffic patterns. This could be used to provide location-based popularity to rank local businesses. For example, to recommend the best local ice cream shop that always has long lines, but happens to be quieter right now. CitySense is similar to predictive traffic routing systems such as Microsoft's ClearFlow system [NY Times]. Eric Horvitz also briefly describes ClearFlow in a recent keynote: Machine Learning, Reasoning, and Intelligence in Daily Life: Directions and Challenges.

As Steve Green recently said, "Recommendation is the new search." An ultimate local recommendation system would utilize many factors including the current time, my location from my mobile, weather, what other people like me have done in the situation in the past, and what other people are doing right now.