E. Charniak and R. Goldman. A Semantics for Probabilistic Quantifier-Free First-Order Languages, with Particular Application to Story Understanding. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence. 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....is common to all the simplest such plans. The technical achievement of this work is the ability to specify the assumptions the recognizer makes without recourse to a control mechanism lying outside the theory. Another natural way to view plan recognition is as a kind of probabilistic reasoning [Charniak Goldman 1989]. The conclusions of the recognizer are simply those statements that are assigned a high probability in light of the evidence. A probabilistic approach is similar to the approach taken in this chapter in that reasoning proceeds directly from the observations to the conclusions and avoids the ....

Charniak, Eugene & Robert Goldman (1989) A Semantics for Probabilistic Quantifier-Free First-Order Languages, with Particular Application to Story Understanding, Proceedings of IJCAI-89, pg. 1074, Morgan-Kaufmann.

.... we can no longer retract the result (monotonic property) Therefore, to understand the phenomena, we need other reasoning methods and in fact, many researches have been using general reasoning frameworks in Artificial Intelligence such as abduction (Hobbs et al. 1993) probabilistic network (Charniak and Goldman, 1989), truth maintenance system (Zernik and Brown, 1988) default logic (Quantz, 1993) and conditional logic (Lascarides, 1993) In this paper, we propose another alternative, that is, circumscription (McCarthy, 1986; Lifschitz, 1985) Even though circumscription is one of the most popular formalisms ....

Charniak, E., and Goldman, R. 1989. A Semantics for Probabilistic Quantifier-Free FirstOrder Languages, with Particular Application to Story Understanding. In Proceedings of IJCAI-89, pages 1074 -- 1079.

....constraints, is robust, and is readily incorporated into standard unification based and frame based models. 1 Introduction The application of Bayesian belief networks (Pearl 1988) to natural language disambiguation problems has recently generated some interest (Goldman Charniak 1990; Charniak Goldman 1988, 1989; Burger Connolly 1992) There is a natural appeal to using the mathematically consistent probability calculus to combine quantitative degrees of evidence for alternative interpretations, and even to help resolve parsing decisions. However, to formulate disambiguation problems using belief ....

CHARNIAK, EUGENE & ROBERT GOLDMAN. 1989. A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of IJCAI-89, Eleventh International Joint Conference on Artificial Intelligence, 1074--1079.

....there has been recent work in the field of machine learning on the automatic construction of belief networks from data. Belief networks have also been used recently in systems for performing visual identification (Binford et al. 1989) and for understanding natural language and identifying plans (Charniak and Goldman 1989). Other Developments The last decade has seen a broad spectrum of advances in automated reasoning methods, beyond techniques for reasoning under uncertainty. I will briefly mention a few areas of research with application to systematic biology. Given the explosion of biomedical literature and ....

Charniak, E. and Goldman, R., A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, Detroit, Michigan. International Joint Conferences on Artificial Intelligence.

....3: Explanation Quality versus Run Time work differs from ours in that unlike their emphasis on mostly linguistics issues like reference resolution and syntactic ambiguity resolution, ACCEL is concerned with constructing deep, causal explanations of the input text. The work of [Charniak, 1986] and [Charniak and Goldman, 1989] are most similar to ours. However, they are primarily concerned with recognizing characters plans and goals in narrative stories, whereas ACCEL is also capable of constructing causal explanations for expository text. For example, a complete understanding of an encyclopedia text describing ....

....of the animals. See [Ng and Mooney, 1989] for a list of expository text sentences that can be processed by ACCEL. Also, explanations are evaluated based on their explanatory coherence in our work, as opposed to the simplicity criterion of [Charniak, 1986] and the probability criterion of [Charniak and Goldman, 1989]. Furthermore, the work of [Charniak, 1986] used marker passing to restrict the search for explanations, whereas we used a form of beam search for the efficient construction of explanations. The Bayesian probabilistic approach to plan recognition and text understanding has been proposed by ....

[Article contains additional citation context not shown here]

Eugene Charniak and Robert P. Goldman. A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, 1989.

....logic of discrete Bayesian Networks, where the logic expresses the object level knowledge and the independence of Bayesian networks is an emergent property of the representation. In the uncertainty community there has also been a need to extend Bayesian networks to beyond a propositional language [5, 6, 22]. Probabilistic Horn abduction is naturally non propositional, and provides a natural extension of Bayesian Probabilistic Horn abduction and Bayesian networks 3 networks to a non propositional language. The work presented in this paper should be contrasted with other attempts to combine logic ....

E. Charniak and R. Goldman. A semantics for probabilistic quantifierfree first-order languages, with particular application to story understanding. In Proc. 11th International Joint Conf. on Artificial Intelligence, pages 1074--1079, Detroit, Mich., August 1989.

....we consider Bayesian networks [16] which is one of the most popular models for uncertainty. Knowledge is organized in a hierarchical graphical fashion providing easy visualization of the reasoning domain. Bayesian networks have seen a wide variety of applications such as story comprehension [4, 10], planning [5, 12, 13] circuit fault detection [16] and medical diagnoses [17] Not surprisingly though, computing with Bayesian networks is NP hard [8] In order to alleviate this computational bottleneck, existing applications and algorithms severely restrict themselves to some smaller class of ....

Eugene Charniak and Robert Goldman. A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of the IJCAI Conference, 1989.

....have probabilistic interpretations as members of this category. Methods of the second category allow compositional constraints about roles and fillers to be predicated, but restrict the precedence of such constraints. The members of this group most relevant here (Goldman Charniak 1990a, 1990b; Charniak Goldman 1988, 1989; Burger Connolly 1992) are based on Bayesian belief networks (Pearl 1988) Difficulties that arise with this group are discussed in a companion paper (Wu 1993a) Briefly, the precedence of dependencies is dictated by the topology of the net, according to the rule that any two nodes are ....

CHARNIAK, EUGENE & ROBERT GOLDMAN. 1989. A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of IJCAI-89, Eleventh International Joint Conferenceon Artificial Intelligence, 1074--1079.

....explicitly but that are given a probabilistic interpretation. The following frameworks are well known ones in artificial intelligence, especially natural language processing. ffl Prolog Extension [59] ffl Marker Passing [48] 9] ffl Cost Based Reasoning [13] 31, 32] ffl Bayesian Network [11, 12, 10] ffl Probabilistic Horn Abduction [55] ffl Fuzzy Logic [17] ffl Inference on Probability [21, 22] Shapiro [59] defined a meta circular interpreter of Prolog which can interpret Prolog programs extended with probabilities, and proposed to prune the derivations that have a smaller probability ....

Eugene Charniak and Robert Goldman. "A Semantics for Probabilistic Quantifier-Free First-Order Languages, with Particular Application to Story Understanding". In Proceedings of the 11th International Joint-Conference on Artificial Intelligence (IJCAI '89), pages 1074--1079, 1989.

....activation. Other notable contributions in text understanding using marker propagations are [10] 5] 6] 14] 21] 1] 29] to mention only a few. Another approach to text inference tried to use the power of predicate logic to implement abductions and other reasoning mechanisms [15] [7], 26] While predicate logic provides precise conclusions when the system has all the information it needs, it fails to provide answers when knowledge is incomplete or uncertain, and unfortunately these are the majority of cases in natural language understanding. It is only recently that ....

E. Charniak and R. Goldman, A semantic for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of the Eleventh International Joint Conference of Artificial Intelligence IJCAI-89, pages 1074--1079, 1989.

....reasoning can provide such an understanding. The three probabilistic plan recognition systems I have looked at make use of two different formalisms for reasoning under uncertainty: one uses Bayesian belief networks [11, 63] and the other two use Dempster Shafer theory [76] The first model [12, 13, 14] addresses the problem of story understanding using Bayesian belief networks to reason about the plans being pursued by characters in the stories. The second model [10] is concerned with incorporating default inferences into reasoning about the plans of a user of a natural language consultation ....

Eugene Charniak and Robert Goldman. A semantics for probabilistic quantifier-free first-order languages. In Proceedings IJCAI-89, pages 1074--1079, Detroit, MI, 1989.

....our web page) The material herein is copyrighted material. It may not be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from AAAI. Articles 50 AI MAGAZINE understanding (Charniak and Goldman 1989a, 1989b; Goldman 1990), vision (Levitt, Mullin, and Binford 1989) heuristic search (Hansson and Mayer 1989) and so on. It is probably fair to say that Bayesian networks are to a large segment of the AI uncertainty community what resolution theorem proving is to the AIlogic community. Nevertheless, despite what seems ....

....the corridor layout (that is, the map) Here, too, the intent is to combine this diagnostic problem with decision theory, so the robot could weigh the alternative of deviating from its planned course to explore portions of the building for which it has no map. My own work on story understanding (Charniak and Goldman 1989a, 1991; Goldman 1990) depends on a similar analogy. called an influence diagram, a concept invented by Howard (Howard and Matheson 1981) In PATHFINDER, decision theory is used to choose the next test to be performed when the current tests are not sufficient to make a diagnosis. PATHFINDER has the ability to make ....

Charniak, E., and Goldman, R. 1989a. A Semantics for Probabilistic Quantifier-Free First-Order Languages with Particular Application to Story Understanding.

....the same distribution for both. We will make virtually identical assumptions for equality and evidence nodes; that is, we will assume that the distribution for a node with a more specific schema is the same as that for the more general. For more detailed discussion of the existential arc, see (Charniak Goldman [1989]) Equality nodes. The distribution of these nodes depends on whether or not an existential bridges the parent nodes; for example, the = 2 node in figure 4.5 has no existential, while = 1 does. In the former case, we want to know the probability that the filler of the store of slot of shopping3 ....

....give the total number of paths returned, with the invalid and low probability paths already weeded out. Second, we counted paths which were asserted , i.e. used for forward chaining and Bayesian network construction. These are paths which passed various secondary filters reported in (Carroll Charniak [1989]) Third, we counted paths whose resulting statements were actually evaluated using our Bayesian network evaluation mechanism. Some paths could be eliminated without evaluation, as we will describe shortly. Finally, we counted those paths which we approved after evaluation; for a path to be ....

Charniak, Eugene & Goldman, Robert P. [1989], "A semantics for probabilistic quantifier-free first-order lanuages, with particular application to story understanding, " IJCAI-89.

....we get further portions of the story. Of course we can reduce the dynamic case to the static. Each scene, medical problem, or story comprehension task sets up a new probabilistic problem which the program tackles afresh. For example, in our previous work on the Wimp3 story understanding program (Charniak Goldman 1989; Charniak Goldman 1991; Goldman Charniak 1990) story decisions are translated into Bayesian Networks. The network is extended on a word by word basis. As one would expect, most of the network at word n was there for word n Gamma 1. Yet the algorithm we used for calculating the probabilities ....

Charniak, E. and Goldman, R. 1989. A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding. In Proceedings of the IJCAI Conference.

No context found.

E. Charniak and R. Goldman. A Semantics for Probabilistic Quantifier-Free First-Order Languages, with Particular Application to Story Understanding. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence. 1989.

No context found.

Charniak, E. and Goldman, R., "A Semantics for Probabilistic Quantifier-Free First-Order Languages with Particular Application to Story Understanding," In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, Menlo Park, California, 1989, pp. 1074-1079.

No context found.

Charniak, Eugene, and Robert Goldman, 1989. "A Semantics for Probabilistic Quantifier-Free First-Order Languages, with Particular Application to Story Understanding ", Proceedings, Eleventh International Joint Conference on Artificial Intelligence, pp. 1074--1079. Detroit, Michigan. August 1989.

No context found.

Charniak, E. and Goldman, R. (1989a) A semantics for probabilistic quantifier-free first-order languages, with particular application to story understanding, IJCAI-89.

No context found.

Charniak, E. and Goldman, R. 1989. A Semantics for Probabilistic Quantifier-Free FirstOrder Languages, with Particular Application to Story Understanding. In Proceedings of the 11th International Joint Conference on Artificial Intelligence, Detroit, MI.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC