Johnson, Mark, Stuart Geman, S. Canon, Z. Chi and S. Riezler (1999). Estimators for Stochastic \Uni cation-based" Grammars. In Proceedings of the ACL, 1999.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....preliminary results, subsequent tests have not enabled us to conclude whether this approach helps the kind of models we are working with. 1 Introduction Recent years have seen a considerable amount of research in the eld of maximum entropy based log linear modeling for disambiguation [1, 7, 9]. This is in large part due to the fact that such models are superior to others in modeling linguistic phenomena which contain internal dependencies, since the log linear modeling framework allows weights to be assigned to features without assuming independence of the features. An important issue ....

....a context to be a sentence and the events within this context are the possible parses of the sentence. Each parse is characterized by a set of feature values, and may be compared on the basis of those features with other possible parses. Parsing is performed as described in section 2.2. Following [9], the best rst search proceeds on the basis of the unnormalized conditional probabilities derived from equation 1 for each possible subtree. 4 The Features and Feature Merging The model depends on the distribution of the features and their informativeness, thus it is important that the features ....

Mark Johnson, Stuart Geman, Stephen Canon, Zhiyi Chi, and Stefan Riezler. Estimators for stochastic \unication-based" grammars. In Proceedings of the 37th Annual Meeting of the ACL, pages 535-541, College Park, Maryland, 1999.

....a variety of parse evaluation functions are described. In the experiments discussed here, the parse evaluation function consisted of a log linear model. Log linear models were introduced to natural language processing by [2] and [11] and applied to stochastic constraint based grammars by [1] and [14]. Given a conditional log linear model, the probability of a sentence x having the parse y is: p(yjx) 1 Z(x) exp X i i f i (x; y) Here, each f i (x; y) is a property function which will return the number of times a speci c property i occurs in parse y of sentence x. Each property ....

Mark Johnson, Stuart Geman, Stephen Canon, Zhiyi Chi, and Stefan Riezler. Estimators for stochastic \unication-based" grammars. In Proceedings of the 37th Annual Meeting of the ACL, pages 535-541, College Park, Maryland, 1999.

....between feature vectors can be computed eciently. We show how this allows SVMs to be applied to representations that would be intractable, or certainly challenging, for other methods. 1 Introduction Recently, methods such as Markov Random Fields (MRFs) Abney 1997; Della Pietra et al. 1997; Johnson et al. 1999) and Boosting (Abney et al. 1999; Collins 2000) have been introduced to natural language problems. These methods have the advantage of being highly exible, in that they allow the objects being modeled to be represented as arbitrary feature vectors. The motivation is that the freedom to use richer ....

.... on NLP tasks: there is empirical evidence for this, in improvements on WSJ treebank parsing through additional features (Charniak 1999; Collins 2000) or in the successful application of MRFs to linguistically motivated representations such as uni cation grammars or LFG parses (Abney 1997; Johnson et al. 1999). Our aim in this paper is to introduce new learning algorithms which take advantage of rich, high dimensional representations of NLP objects. Cortes Vapnik 1995) describe two major problems when working with such rich representations. The rst is technical : how can models be trained and ....

[Article contains additional citation context not shown here]

Johnson, M., Geman, S., Canon, S., Chi, S., & Riezler, S. (1999). Estimators for stochastic `unicationbased " grammars. In Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics.

No context found.

Johnson, Mark, Stuart Geman, S. Canon, Z. Chi and S. Riezler (1999). Estimators for Stochastic \Uni cation-based" Grammars. In Proceedings of the ACL, 1999.

No context found.

M. Johnson, S. Geman, S. Canon, Z. Chi, and S. Riezler. Estimators for stochastic \unication-based" grammars. In Proc. of ACL'99, 1999.

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