Results 1  10
of
30
Grounding the lexical semantics of verbs in visual perception using force dynamics and event logic
 Journal of Artificial Intelligence Research
, 2001
"... This paper presents an implemented system for recognizing the occurrence of events described by simple spatialmotion verbs in short image sequences. The semantics of these verbs is specified with eventlogic expressions that describe changes in the state of forcedynamic relations between the parti ..."
Abstract

Cited by 95 (2 self)
 Add to MetaCart
This paper presents an implemented system for recognizing the occurrence of events described by simple spatialmotion verbs in short image sequences. The semantics of these verbs is specified with eventlogic expressions that describe changes in the state of forcedynamic relations between the participants of the event. An efficient finite representation is introduced for the infinite sets of intervals that occur when describing liquid and semiliquid events. Additionally, an efficient procedure using this representation is presented for inferring occurrences of compound events, described with eventlogic expressions, from occurrences of primitive events. Using force dynamics and event logic to specify the lexical semantics of events allows the system to be more robust than prior systems based on motion profile. 1.
Multiple representations of knowledge in a mechanics problem solver
 In Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI77
, 1977
"... Expert problemsolving programs have focused on working problems which humans consider difficult Oddly, many such problemsolvers could not solve less difficult versions of the problems addressed by their expertise This shortcoming also contributed to these programs ' inability to solve harder probl ..."
Abstract

Cited by 52 (0 self)
 Add to MetaCart
Expert problemsolving programs have focused on working problems which humans consider difficult Oddly, many such problemsolvers could not solve less difficult versions of the problems addressed by their expertise This shortcoming also contributed to these programs ' inability to solve harder problems. To overcome this 'paradox ' requires multiple representations of knowledge, inferencing schemes for each, and communication schemes between them. This paper presents a program, NEWTON, applying this idea to the domain of simple classical mechanics. NEWTON employs the method of envisionment, whereby simple questions may be answered directly, and plans produced for solving more complex problems. Envisioning enables NEWTON to use qualitative arguments when possible, with resorts to mathematical equations only if the qualitative reasoning fails to produce a solution.
Towards robust semantic role labeling
 Computational Linguistics
, 2008
"... Most research on semantic role labeling (SRL) has been focused on training and evaluating on the same corpus in order to develop the technology. This strategy, while appropriate for initiating research, can lead to overtraining to the particular corpus. The work presented in this paper focuses on a ..."
Abstract

Cited by 26 (2 self)
 Add to MetaCart
Most research on semantic role labeling (SRL) has been focused on training and evaluating on the same corpus in order to develop the technology. This strategy, while appropriate for initiating research, can lead to overtraining to the particular corpus. The work presented in this paper focuses on analyzing the robustness of an SRL system when trained on one genre of data and used to label a different genre. Our stateoftheart semantic role labeling system, while performing well on WSJ test data, shows significant performance degradation when applied to data from the Brown corpus. We present a series of experiments designed to investigate the source of this lack of portability. These experiments are based on comparisons of performance using PropBanked WSJ data and PropBanked Brown corpus data. Our results indicate that while syntactic parses and argument identification port relatively well to a new genre, argument classification does not. Our analysis of the reasons for this is presented and generally point to the nature of the more lexical/semantic features dominating the classification task and general structural features dominating the argument identification task. 1
Representations of Knowledge in a Program for Solving Physics Problems
 In Proceedings of the Fifth International Joint Conference on Artificial Intelligence
, 1977
"... A computer program which solves physics problems stated in English is described in terms of the knowledge which is used to transform one type of representation into another. The English sentences of the problem statement are progressively transformed into a semantic network form, a languagefree int ..."
Abstract

Cited by 17 (1 self)
 Add to MetaCart
A computer program which solves physics problems stated in English is described in terms of the knowledge which is used to transform one type of representation into another. The English sentences of the problem statement are progressively transformed into a semantic network form, a languagefree internal model of the objects in the problem and their attributes and relationships, a set of canonical object frames which interpret actual objects as canonical objects (such as a point mass) , a geometric model, a set of equations, and a picture model. The general notion of a canonical object frame, which abstracts a subset of the properties of an object to form a representation of a canonical object whose interactions with related canonical objects can be formally modelled, is discussed as a method of organizing problemsolving programs. 1.
Statistical Source Channel Models for Natural Language Understanding
, 1996
"... d my ignorance in the field. He was always patient, and took the time to explain his answers at a level I could understand. iv Dr. Todd Ward, a colleague of mine at IBM, has also "been there" for me. I cannot count the number of times that Todd helped me figure out a solution to a problem, either ..."
Abstract

Cited by 10 (1 self)
 Add to MetaCart
d my ignorance in the field. He was always patient, and took the time to explain his answers at a level I could understand. iv Dr. Todd Ward, a colleague of mine at IBM, has also "been there" for me. I cannot count the number of times that Todd helped me figure out a solution to a problem, either mathematical or programming. Whenever I was not sure about a solution to a problem, Todd was my sounding board. I'm sure that his individual research efforts were slowed by our meetings, but that never stopped him from helping me. Todd also acted as a counselor, providing insight on how to complete a doctorate! Former IBMer, Dr. Stephen Della Pietra, is without a doubt the brightest mathematician with whom I have ever worked. Like Salim and Todd, he knows statistical modeling at a much greater depth than I do, and he never minded "bringing down" the level of his explanations to one where I could understand and absorb the material. Stephen was my mentor, and without his expert tutelag
Future Directions of Machine Translation
, 1986
"... this paper, we will discuss several problems concerned with 'understanding and translation', especially how we can integrate the two lines of research, with their different histories and different techniques, into unified frameworks, and the difficulties we might encounter in attempting such an inte ..."
Abstract

Cited by 9 (0 self)
 Add to MetaCart
this paper, we will discuss several problems concerned with 'understanding and translation', especially how we can integrate the two lines of research, with their different histories and different techniques, into unified frameworks, and the difficulties we might encounter in attempting such an integration. The discussion wil] reveal some of the reasons why MT researches are so separated from the research in tile other application fields of NLP. We will also list some of the key problems, both linguistic and computational, which we encountered during the development of our MT systems, the Mu systems INsgee 1984, 85, 86] [Tsujii 1984, 85] [Nakamura 1984, 86 ] [Sakamoto 1984], and whose resolutions we consider to be of essential importance for future MT research and development
Understanding Mathematical Discourse
 Dialogue. Amsterdam University
, 1999
"... Discourse Understanding is hard. This seems to be especially true for mathematical discourse, that is proofs. Restricting discourse to mathematical discourse allow us, however, to study the subject matter in its purest form. This domain of discourse is rich and welldefined, highly structured, offers ..."
Abstract

Cited by 7 (6 self)
 Add to MetaCart
Discourse Understanding is hard. This seems to be especially true for mathematical discourse, that is proofs. Restricting discourse to mathematical discourse allow us, however, to study the subject matter in its purest form. This domain of discourse is rich and welldefined, highly structured, offers a welldefined set of discourse relations and forces/allows us to apply mathematical reasoning. We give a brief discussion on selected linguistic phenomena of mathematical discourse, and an analysis from the mathematician’s point of view. Requirements for a theory of discourse representation are given, followed by a discussion of proofs plans that provide necessary context and structure. A large part of semantics construction is defined in terms of proof plan recognition and instantiation by matching and attaching. 1
Seeing language learning inside the math: Cognitive analysis yields transfer
 Cognitive Science Society
, 2010
"... Achieving and understanding effective transfer of learning requires a careful analysis of the hidden knowledge and skills to be transferred. We present an experiment that tests a subtle prediction of such an analysis. It concluded that a critical difficulty in students ’ learning to translate algebr ..."
Abstract

Cited by 6 (6 self)
 Add to MetaCart
Achieving and understanding effective transfer of learning requires a careful analysis of the hidden knowledge and skills to be transferred. We present an experiment that tests a subtle prediction of such an analysis. It concluded that a critical difficulty in students ’ learning to translate algebra story problems into symbolic expressions is in learning the grammar of such expressions. We hypothesized that exercises requiring students to substitute one algebraic expression into another would enhance students ’ algebraic grammar knowledge. This hypothesis led to a counterintuitive prediction that learning to symbolize story problems could be better enhanced through practice on dissimilar looking substitution exercises than through practice on more similar looking story problems. We report on an experimental comparison involving 303 middle school students that supports this prediction. We discuss how having learners externalize a uniform abstract form and get interactive feedback on it may be important factors in enhancing transfer.
Verifying Textbook Proofs
 INT. WORKSHOP ON FIRSTORDER THEOREM PROVING (FTP'98), TECHNICAL REPORT E1852GS981
, 1998
"... ..."
Toward a Dynamic Model of Early Algebra Acquisition
 In Proceedings of the European Conference on AI in Education
, 1996
"... Abstract: How does one go about creating quality cognitive models that capture the difficulties students have in learning complex skills? The answer we propose is to break a domain down into a number of dimensions (or difficulty factors) and use cognitive modeling and empirical work to better unders ..."
Abstract

Cited by 3 (2 self)
 Add to MetaCart
Abstract: How does one go about creating quality cognitive models that capture the difficulties students have in learning complex skills? The answer we propose is to break a domain down into a number of dimensions (or difficulty factors) and use cognitive modeling and empirical work to better understand them. We demonstrate how this can be done for the domain of algebra. This type of analysis can not only lead to a better understanding of the domain for traditional instruction, but it can also serve as the foundation for the development of computer tutors.