Our goal is to develop general-purpose techniques for probabilistic reasoning and learning in structured spaces. These spaces are characterized by complex logical constraints on what constitutes a possible world. We propose a tractable formalism, called probabilistic sentential decision diagrams, and show it effectively learns structured probability distributions in two applications: product configuration and preference learning.

Relational learning can be described as the task of learning first-order logic rules from examples. It has enabled a number of new machine learning applications, e.g. graph mining and link analysis in social networks. The CILP++ system is a neural-symbolic system which can perform efficient relational learning, by being able to process first-order logic knowledge into a neural network. CILP++ relies on BCP, a recently discovered propositionalization algorithm, to perform relational learning. However, efficient knowledge extraction from such networks is an open issue and features generated by BCP do not have an independent relational description, which prevents sound knowledge extraction from such networks. We present a methodology for generating independent propositional features for BCP by using semi-propositionalization of bottom clauses. Empirical results obtained in comparison with the original version of BCP show that this approach has comparable accuracy and runtimes, while allowing proper relational knowledge representation of features for knowledge extraction from CILP++ networks.

In order to be fully robust and responsive to a dynamically changing real-world environment, intelligent robots will need to engage in a variety of simultaneous reasoning modalities. In particular, in this paper we consider their needs to i) reason with commonsense knowledge, ii) model their nondeterministic action outcomes and partial observability, and iii) plan toward maximizing long-term rewards. On one hand, Answer Set Programming (ASP) is good at representing and reasoning with commonsense and default knowledge, but is ill-equipped to plan under probabilistic uncertainty. On the other hand, Partially Observable Markov Decision Processes (POMDPs) are strong at planning under uncertainty toward maximizing long-term rewards, but are not designed to incorporate commonsense knowledge and inference. This paper introduces the CORPP algorithm which combines P-log, a probabilistic extension of ASP, with POMDPs to integrate commonsense reasoning with planning under uncertainty. Our approach is fully implemented and tested on a shopping request identification problem both in simulation and on a real robot. Compared with existing approaches using P-log or POMDPs individually, we observe significant improvements in both efficiency and accuracy.

In this paper we assess our progress toward creating a virtual human negotiation agent with fluid turn-taking skills. To facilitate the design of this agent, we have collected a corpus of human-human negotiation roleplays as well as a corpus of Wizard-controlled human-agent negotiations in the same roleplay scenario. We compare the natural turn-taking behavior in our human-human corpus with that achieved in our Wizard-of-Oz corpus, and quantify our virtual human's turn-taking skills using a combination of subjective and objective metrics. We also discuss our design for a Wizard user interface to support real-time control of the virtual human's turn-taking and dialogue behavior, and analyze our wizard's usage of this interface.

A physically situated robot in conversational situations, which is enabled to understand and generate human conversational protocols, has a big potential to facilitate conversation in a small group. In this paper, we present a computational model of facilitation process in a small group, including (1) procedural behaviour decision process controlling engage- ment density to regulate a socially imbalanced situation and (2) language generation process associated with user models, which attempts to trigger participants’ interests. We implemented the model on a conversational robot and assessed the effectiveness of the procedural behavior generation.

For humans, learning abstract concepts and learning language  go hand in hand: we acquire abstract knowledge primarily through  linguistic experience, and acquiring abstract concepts is a crucial  step in learning the meanings of linguistic expressions. Number  knowledge is a case in point: we largely acquire concepts such as  seventy-three through linguistic means, and we can only know what  the sentence ``seventy-three is more than twice as big as  thirty-one" means if we can grasp the meanings of its component  number words. How do we begin to solve this problem? One approach is  to estimate the distribution from which sentences are drawn, and, in  doing so, infer the latent concepts and relationships that best  explain those sentences. We present early work on a learning  framework called Latent Predicate Networks (LPNs) which learns  concepts by inferring the parameters of probabilistic  context-sensitive grammars over sentences.  We show that for a small  fragment of sentences expressing relationships between English  number words, we can use hierarchical Bayesian inference to learn  grammars that can answer simple queries about previously unseen  relationships within this domain. These generalizations demonstrate  LPNs' promise as a tool for learning and representing conceptual  knowledge in language.

The accompanying video captures the multi-modal data displays and speech dialogue of a human Commander (C) and a human Robot Navigator (RN) tele-operating a mobile robot (R) in a remote, previously unexplored area. We describe unique challenges for automation of turn-taking and coordination processes observed in the data.

A formal framework is given for the postulate characterizability of a class of belief revision operators, obtained from a class of partial preorders using minimization. It is shown that for classes of posets characterizability is equivalent to a special kind of definability in monadic second-order logic, which turns out to be incomparable to first-order definability. Several examples are given of characterizable and non-characterizable classes. For example, it is shown that the class of revision operators obtained from posets which are not total is not characterizable.

We have recently shown how to combine random walk inference over knowledge bases with vector space representations of surface forms, improving performance on knowledge base inference. In this paper, we formalize the connection of our prior work to logical inference rules, giving some general observations about methods for incorporating vector space representations into symbolic logic systems. Additionally, we present some promising preliminary work that extends these techniques to learning open-domain relations for the purpose of answering multiple choice questions, achieving 67% accuracy on a small test set.

The design of ontologies for new commonsense domains continues to pose challenges, particularly in cases where multiple potential axiomatizations satisfy the requirements for the ontology. One approach is to specify the requirements with respect to the intended semantics of the terminology; from a mathematical perspective the requirements may be characterized by the class of structures(referred to as the required models) which capturethe intended semantics. This approach leads to a natural notion of the correctness as a relationship between the models of the axiomatization of the ontology and the required models for the ontology. In this paper, we consider three possible generalizations of the notion of the correctness of an ontology in the case in which the ontology and the required models have different signatures.We show that these notions of correctness lead to different approaches for ontology evaluation and discuss the benefits and drawbacks of each approach.