Doug Riecken

An architecture of diversity for commonsense reasoning

Although computers excel at certain bounded tasks that are difficult for humans, such as solving integrals, they have difficulty performing commonsense tasks that are easy for humans, such as understanding stories. In this Technical Forum contribution, we discuss commonsense reasoning and what makes it difficult for computers. We contend that commonsense reasoning is too hard a problem to solve using any single artificial intelligence technique. We propose a multilevel architecture consisting of diverse reasoning and representation techniques that collaborate and reflect in order to allow the best techniques to be used for the many situations that arise in commonsense reasoning. We present story understanding—specifically, understanding and answering questions about progressively harder children’s texts—as a task for evaluating and scaling up a commonsense reasoning system.

Personalized communication networks

COMMUNICATIONS OF THE ACM August 2000/Vol. 43, No. 8 41 DR: It appears the Internet has been a motivating force in the design and creation of service features that attempt to personalize and help users and customers satisfy their individual goals in the use of many Web services. Do you concur with this point of view? KE: I believe it has. I think we are at the beginning of what the Internet and personalization are going to enable users and customers to do. It is important to note that personalization means many things to many people. Generally, what it means to me is oneto-one anything, even though it’s a one-to-many experience. So, if you think about this as the Web, anybody can be going to the Web, but it’s when I interact with the Web that it’s a personalized experience to me. So a useful question is, how do you actually enable my user experience to be something that’s enriching, compelling, serves my needs, and target goal alignment in terms of fulfilling my needs? I think the Internet with its Web services is uniquely positioned to really enable personalization because—I hate to use this old phrase—“its the unifying protocol between computing and communication.” Prior to this, we were able to do personalized communication through the telephone or personalized email, both of which are a one-to-one or one-to-many experience. It is not until you start to marry all the applications imaginable within the industry and then marry that with some front end—call it Web veneer, artificial intelligence, or a type of search engine—that a service then feels personal to me. It’s when I can put in filters, agents, personal screens, some set of rules defining what kind of advertising I want to receive, that I have personal control over the flow of information I send and receive. I think the industry has a tremendous ramp ahead of it over the next 5 to 10 years, but the benefits will be rewarding for everyone. I suspect this might be the last major platform for computing in my lifetime. That does not mean there will not be another one beyond it.

Personal end-user tools

COMMUNICATIONS OF THE ACM August 2000/Vol. 43, No. 8 89 I wish to consider the potential of enduser tools as a next step in the evolution of personalization. What if end users could use their own personal specialized tools to perform tasks anytime, anywhere? What if I had a better bookmark tool than the one embedded in my Web browser? I suspect I would prefer to use my standalone bookmark tool with the browser. Let’s consider another situation. We often use tools embedded in the various content obtained from services on the Internet. For example, a given Web page may contain a calculator tool to compute mortgage rates, but once again, what if I have a better software mortgage calculator tool? I would prefer to use my personal calculator in real time with the Web service instead of the embedded tool provided by the service. Let me catch myself here, I wish to point out that this focus of end-user tools is not restricted to Webbased services. As a matter of fact, why can’t I have a universal bookmark tool that works with my unified messaging service, with my screen-me reach-me service, with my cell phone during actual conversations, with my cable entertainment service, and a range of other services and activities? Right now you might be asking, what is a universal bookmark? Well, that’s the vision and research. The main idea here is about empowering real end users with their own tools to do what they want or need to do the way they would like to do it! End users enabled to perform a task with their own personal tool that allows them to perform a task in the manner they wish to achieve a specific desired result might surely exemplify a very personalized opportunity. In the near future, end users will perform tasks with their own software tools. A number of these tools will not be embedded within a given application, operating system, or service, but instead will exist as personal, individual software tools.

Constraining model-based reasoning using contexts

Web-based customer service has become a norm of business practice with increasing emphasis on modeling customer needs and providing them with targeted or personalized service solutions in a timely fashion. Almost all the commercial Web service systems adopt some kind of simple customer segmentation models and shallow pattern matching or rule-based techniques for high performance. The models built based on these techniques though very efficient have a fundamental limitation in their ability to capture and explain the reasoning in the process of determining and selecting appropriate services or products. However, using deep models (e.g. semantic networks), though desirable for their expressive power, may require significantly more computational resources (e.g. time) for reasoning. This can compromise the system performance. We report on a new approach that represents and uses contextual information in semantic net-based models to constrain and prune potentially very large search space, which results in much improved performance in terms of speed and selectivity as evidenced by the evaluation results.

In Honor of Marvin Minsky's Contributions on his 80th Birthday

Marvin Lee Minsky, a founder of the field of artificial intelligence and professor at MIT, celebrated his 80th birthday on August 9, 2007. This article seizes an opportune time to honor Marvin and his contributions and influence in artificial intelligence, science, and beyond. The article provides readers with some personal insights of Minsky from Danny Hillis, John McCarthy, Tom Mitchell, Erik Mueller, Doug Riecken, Aaron Sloman, and Patrick Henry Winston -- all members of the AI community that Minsky helped to found. The article continues with a brief resume of Minskys research, which spans an enormous range of fields. It concludes with a short biographical account of Minskys personal history.

A commonsense opportunity for computing

There are certainly other technical approaches. For example, we can build fault-tolerant systems out of unreliable components. Is there a way to do the same for security? While that might improve the odds, it is unlikely to provide a perfect security shield. Fault-tolerant systems deal with natural failures, and nature (Einstein reminded us) is subtle but not malicious. Hackers do their best to shift the odds and to create improbable situations they can exploit. If we succeed at this challenge—if we can build distributed systems and a cyber society that is attackresistant—then our networks should survive and even flourish. No one expects major cities to be 100% crime-free, but we do expect to be able to carry out our daily activities with a reasonable degree of safety. The same can and should be true of the Net. There will never be absolute safety and perfect assurance, online or off—but there never was. The Vandals became vandals, and descended to slashing car tires. Today they are hackers, and deface Web sites. We must ensure they do not become Hackers and destroy our cities, or even our enjoyment of them.