id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
cmp-lg/9708013
explanation-based learning of data oriented parsing
cmp-lg cs.CL
This paper presents a new view of Explanation-Based Learning (EBL) of natural language parsing. Rather than employing EBL for specializing parsers by inferring new ones, this paper suggests employing EBL for learning how to reduce ambiguity only partially. The present method consists of an EBL algorithm for learning partial-parsers, and a parsing algorithm which combines partial-parsers with existing ``full-parsers". The learned partial-parsers, implementable as Cascades of Finite State Transducers (CFSTs), recognize and combine constituents efficiently, prohibiting spurious overgeneration. The parsing algorithm combines a learned partial-parser with a given full-parser such that the role of the full-parser is limited to combining the constituents, recognized by the partial-parser, and to recognizing unrecognized portions of the input sentence. Besides the reduction of the parse-space prior to disambiguation, the present method provides a way for refining existing disambiguation models that learn stochastic grammars from tree-banks. We exhibit encouraging empirical results using a pilot implementation: parse-space is reduced substantially with minimal loss of coverage. The speedup gain for disambiguation models is exemplified by experiments with the DOP model.
cmp-lg/9709001
The Complexity of Recognition of Linguistically Adequate Dependency Grammars
cmp-lg cs.CL
Results of computational complexity exist for a wide range of phrase structure-based grammar formalisms, while there is an apparent lack of such results for dependency-based formalisms. We here adapt a result on the complexity of ID/LP-grammars to the dependency framework. Contrary to previous studies on heavily restricted dependency grammars, we prove that recognition (and thus, parsing) of linguistically adequate dependency grammars is NP-complete.
cmp-lg/9709002
Learning Methods for Combining Linguistic Indicators to Classify Verbs
cmp-lg cs.CL
Fourteen linguistically-motivated numerical indicators are evaluated for their ability to categorize verbs as either states or events. The values for each indicator are computed automatically across a corpus of text. To improve classification performance, machine learning techniques are employed to combine multiple indicators. Three machine learning methods are compared for this task: decision tree induction, a genetic algorithm, and log-linear regression.
cmp-lg/9709003
Combining Multiple Methods for the Automatic Construction of Multilingual WordNets
cmp-lg cs.CL
This paper explores the automatic construction of a multilingual Lexical Knowledge Base from preexisting lexical resources. First, a set of automatic and complementary techniques for linking Spanish words collected from monolingual and bilingual MRDs to English WordNet synsets are described. Second, we show how resulting data provided by each method is then combined to produce a preliminary version of a Spanish WordNet with an accuracy over 85%. The application of these combinations results on an increment of the extracted connexions of a 40% without losing accuracy. Both coarse-grained (class level) and fine-grained (synset assignment level) confidence ratios are used and evaluated. Finally, the results for the whole process are presented.
cmp-lg/9709004
Integrating a Lexical Database and a Training Collection for Text Categorization
cmp-lg cs.CL
Automatic text categorization is a complex and useful task for many natural language processing applications. Recent approaches to text categorization focus more on algorithms than on resources involved in this operation. In contrast to this trend, we present an approach based on the integration of widely available resources as lexical databases and training collections to overcome current limitations of the task. Our approach makes use of WordNet synonymy information to increase evidence for bad trained categories. When testing a direct categorization, a WordNet based one, a training algorithm, and our integrated approach, the latter exhibits a better perfomance than any of the others. Incidentally, WordNet based approach perfomance is comparable with the training approach one.
cmp-lg/9709005
A generation algorithm for f-structure representations
cmp-lg cs.CL
This paper shows that previously reported generation algorithms run into problems when dealing with f-structure representations. A generation algorithm that is suitable for this type of representations is presented: the Semantic Kernel Generation (SKG) algorithm. The SKG method has the same processing strategy as the Semantic Head Driven generation (SHDG) algorithm and relies on the assumption that it is possible to compute the Semantic Kernel (SK) and non Semantic Kernel (Non-SK) information for each input structure.
cmp-lg/9709006
Semantic Processing of Out-Of-Vocabulary Words in a Spoken Dialogue System
cmp-lg cs.CL
One of the most important causes of failure in spoken dialogue systems is usually neglected: the problem of words that are not covered by the system's vocabulary (out-of-vocabulary or OOV words). In this paper a methodology is described for the detection, classification and processing of OOV words in an automatic train timetable information system. The various extensions that had to be effected on the different modules of the system are reported, resulting in the design of appropriate dialogue strategies, as are encouraging evaluation results on the new versions of the word recogniser and the linguistic processor.
cmp-lg/9709007
Using WordNet to Complement Training Information in Text Categorization
cmp-lg cs.CL
Automatic Text Categorization (TC) is a complex and useful task for many natural language applications, and is usually performed through the use of a set of manually classified documents, a training collection. We suggest the utilization of additional resources like lexical databases to increase the amount of information that TC systems make use of, and thus, to improve their performance. Our approach integrates WordNet information with two training approaches through the Vector Space Model. The training approaches we test are the Rocchio (relevance feedback) and the Widrow-Hoff (machine learning) algorithms. Results obtained from evaluation show that the integration of WordNet clearly outperforms training approaches, and that an integrated technique can effectively address the classification of low frequency categories.
cmp-lg/9709008
Semantic Similarity Based on Corpus Statistics and Lexical Taxonomy
cmp-lg cs.CL
This paper presents a new approach for measuring semantic similarity/distance between words and concepts. It combines a lexical taxonomy structure with corpus statistical information so that the semantic distance between nodes in the semantic space constructed by the taxonomy can be better quantified with the computational evidence derived from a distributional analysis of corpus data. Specifically, the proposed measure is a combined approach that inherits the edge-based approach of the edge counting scheme, which is then enhanced by the node-based approach of the information content calculation. When tested on a common data set of word pair similarity ratings, the proposed approach outperforms other computational models. It gives the highest correlation value (r = 0.828) with a benchmark based on human similarity judgements, whereas an upper bound (r = 0.885) is observed when human subjects replicate the same task.
cmp-lg/9709009
Evaluating Parsing Schemes with Entropy Indicators
cmp-lg cs.CL
This paper introduces an objective metric for evaluating a parsing scheme. It is based on Shannon's original work with letter sequences, which can be extended to part-of-speech tag sequences. It is shown that this regular language is an inadequate model for natural language, but a representation is used that models language slightly higher in the Chomsky hierarchy. We show how the entropy of parsed and unparsed sentences can be measured. If the entropy of the parsed sentence is lower, this indicates that some of the structure of the language has been captured. We apply this entropy indicator to support one particular parsing scheme that effects a top down segmentation. This approach could be used to decompose the parsing task into computationally more tractable subtasks. It also lends itself to the extraction of predicate/argument structure.
cmp-lg/9709010
Message-Passing Protocols for Real-World Parsing -- An Object-Oriented Model and its Preliminary Evaluation
cmp-lg cs.CL
We argue for a performance-based design of natural language grammars and their associated parsers in order to meet the constraints imposed by real-world NLP. Our approach incorporates declarative and procedural knowledge about language and language use within an object-oriented specification framework. We discuss several message-passing protocols for parsing and provide reasons for sacrificing completeness of the parse in favor of efficiency based on a preliminary empirical evaluation.
cmp-lg/9709011
Off-line Parsability and the Well-foundedness of Subsumption
cmp-lg cs.CL
Typed feature structures are used extensively for the specification of linguistic information in many formalisms. The subsumption relation orders TFSs by their information content. We prove that subsumption of acyclic TFSs is well-founded, whereas in the presence of cycles general TFS subsumption is not well-founded. We show an application of this result for parsing, where the well-foundedness of subsumption is used to guarantee termination for grammars that are off-line parsable. We define a new version of off-line parsability that is less strict than the existing one; thus termination is guaranteed for parsing with a larger set of grammars.
cmp-lg/9709012
Using Single Layer Networks for Discrete, Sequential Data: An Example from Natural Language Processing
cmp-lg cs.CL
A natural language parser which has been successfully implemented is described. This is a hybrid system, in which neural networks operate within a rule based framework. It can be accessed via telnet for users to try on their own text. (For details, contact the author.) Tested on technical manuals, the parser finds the subject and head of the subject in over 90% of declarative sentences. The neural processing components belong to the class of Generalized Single Layer Networks (GSLN). In general, supervised, feed-forward networks need more than one layer to process data. However, in some cases data can be pre-processed with a non-linear transformation, and then presented in a linearly separable form for subsequent processing by a single layer net. Such networks offer advantages of functional transparency and operational speed. For our parser, the initial stage of processing maps linguistic data onto a higher order representation, which can then be analysed by a single layer network. This transformation is supported by information theoretic analysis.
cmp-lg/9709013
An Abstract Machine for Unification Grammars
cmp-lg cs.CL
This work describes the design and implementation of an abstract machine, Amalia, for the linguistic formalism ALE, which is based on typed feature structures. This formalism is one of the most widely accepted in computational linguistics and has been used for designing grammars in various linguistic theories, most notably HPSG. Amalia is composed of data structures and a set of instructions, augmented by a compiler from the grammatical formalism to the abstract instructions, and a (portable) interpreter of the abstract instructions. The effect of each instruction is defined using a low-level language that can be executed on ordinary hardware. The advantages of the abstract machine approach are twofold. From a theoretical point of view, the abstract machine gives a well-defined operational semantics to the grammatical formalism. This ensures that grammars specified using our system are endowed with well defined meaning. It enables, for example, to formally verify the correctness of a compiler for HPSG, given an independent definition. From a practical point of view, Amalia is the first system that employs a direct compilation scheme for unification grammars that are based on typed feature structures. The use of amalia results in a much improved performance over existing systems. In order to test the machine on a realistic application, we have developed a small-scale, HPSG-based grammar for a fragment of the Hebrew language, using Amalia as the development platform. This is the first application of HPSG to a Semitic language.
cmp-lg/9709014
Amalia -- A Unified Platform for Parsing and Generation
cmp-lg cs.CL
Contemporary linguistic theories (in particular, HPSG) are declarative in nature: they specify constraints on permissible structures, not how such structures are to be computed. Grammars designed under such theories are, therefore, suitable for both parsing and generation. However, practical implementations of such theories don't usually support bidirectional processing of grammars. We present a grammar development system that includes a compiler of grammars (for parsing and generation) to abstract machine instructions, and an interpreter for the abstract machine language. The generation compiler inverts input grammars (designed for parsing) to a form more suitable for generation. The compiled grammars are then executed by the interpreter using one control strategy, regardless of whether the grammar is the original or the inverted version. We thus obtain a unified, efficient platform for developing reversible grammars.
cmp-lg/9709015
Segmentation of Expository Texts by Hierarchical Agglomerative Clustering
cmp-lg cs.CL
We propose a method for segmentation of expository texts based on hierarchical agglomerative clustering. The method uses paragraphs as the basic segments for identifying hierarchical discourse structure in the text, applying lexical similarity between them as the proximity test. Linear segmentation can be induced from the identified structure through application of two simple rules. However the hierarchy can be used also for intelligent exploration of the text. The proposed segmentation algorithm is evaluated against an accepted linear segmentation method and shows comparable results.
cmp-lg/9710001
Use of Weighted Finite State Transducers in Part of Speech Tagging
cmp-lg cs.CL
This paper addresses issues in part of speech disambiguation using finite-state transducers and presents two main contributions to the field. One of them is the use of finite-state machines for part of speech tagging. Linguistic and statistical information is represented in terms of weights on transitions in weighted finite-state transducers. Another contribution is the successful combination of techniques -- linguistic and statistical -- for word disambiguation, compounded with the notion of word classes.
cmp-lg/9710002
Tagging French Without Lexical Probabilities -- Combining Linguistic Knowledge And Statistical Learning
cmp-lg cs.CL
This paper explores morpho-syntactic ambiguities for French to develop a strategy for part-of-speech disambiguation that a) reflects the complexity of French as an inflected language, b) optimizes the estimation of probabilities, c) allows the user flexibility in choosing a tagset. The problem in extracting lexical probabilities from a limited training corpus is that the statistical model may not necessarily represent the use of a particular word in a particular context. In a highly morphologically inflected language, this argument is particularly serious since a word can be tagged with a large number of parts of speech. Due to the lack of sufficient training data, we argue against estimating lexical probabilities to disambiguate parts of speech in unrestricted texts. Instead, we use the strength of contextual probabilities along with a feature we call ``genotype'', a set of tags associated with a word. Using this knowledge, we have built a part-of-speech tagger that combines linguistic and statistical approaches: contextual information is disambiguated by linguistic rules and n-gram probabilities on parts of speech only are estimated in order to disambiguate the remaining ambiguous tags.
cmp-lg/9710003
Disambiguating with Controlled Disjunctions
cmp-lg cs.CL
In this paper, we propose a disambiguating technique called controlled disjunctions. This extension of the so-called named disjunctions relies on the relations existing between feature values (covariation, control, etc.). We show that controlled disjunctions can implement different kind of ambiguities in a consistent and homogeneous way. We describe the integration of controlled disjunctions into a HPSG feature structure representation. Finally, we present a direct implementation by means of delayed evaluation and we develop an example within the functionnal programming paradigm.
cmp-lg/9710004
Parsing syllables: modeling OT computationally
cmp-lg cs.CL
In this paper, I propose to implement syllabification in OT as a parser. I propose several innovations that result in a finite and small candidate set. The candidate set problem is handled with several moves: i) MAX and DEP violations are not hypothesized by the parser, ii) candidates are encoded locally, and iii) EVAL is applied constraint by constraint. The parser I propose is implemented in Prolog. It has a number of desirable consequences. First, it runs and thus provides an existence proof that syllabification can be implemented in OT. There are a number of other desirable consequences as well. First, constraints are implemented as finite-state transducers. Second, the parser makes several interesting claims about the phonological properties of so-called nonrecoverable insertions and deletions. Third, the implementation suggests some particular reformulations of some of the benchmark constraints in the OT arsenal, e.g. *COMPLEX, PARSE, ONSET, and NOCODA.
cmp-lg/9710005
Attaching Multiple Prepositional Phrases: Generalized Backed-off Estimation
cmp-lg cs.CL
There has recently been considerable interest in the use of lexically-based statistical techniques to resolve prepositional phrase attachments. To our knowledge, however, these investigations have only considered the problem of attaching the first PP, i.e., in a [V NP PP] configuration. In this paper, we consider one technique which has been successfully applied to this problem, backed-off estimation, and demonstrate how it can be extended to deal with the problem of multiple PP attachment. The multiple PP attachment introduces two related problems: sparser data (since multiple PPs are naturally rarer), and greater syntactic ambiguity (more attachment configurations which must be distinguished). We present and algorithm which solves this problem through re-use of the relatively rich data obtained from first PP training, in resolving subsequent PP attachments.
cmp-lg/9710006
Learning Features that Predict Cue Usage
cmp-lg cs.CL
Our goal is to identify the features that predict the occurrence and placement of discourse cues in tutorial explanations in order to aid in the automatic generation of explanations. Previous attempts to devise rules for text generation were based on intuition or small numbers of constructed examples. We apply a machine learning program, C4.5, to induce decision trees for cue occurrence and placement from a corpus of data coded for a variety of features previously thought to affect cue usage. Our experiments enable us to identify the features with most predictive power, and show that machine learning can be used to induce decision trees useful for text generation.
cmp-lg/9710007
A Corpus-Based Investigation of Definite Description Use
cmp-lg cs.CL
We present the results of a study of definite descriptions use in written texts aimed at assessing the feasibility of annotating corpora with information about definite description interpretation. We ran two experiments, in which subjects were asked to classify the uses of definite descriptions in a corpus of 33 newspaper articles, containing a total of 1412 definite descriptions. We measured the agreement among annotators about the classes assigned to definite descriptions, as well as the agreement about the antecedent assigned to those definites that the annotators classified as being related to an antecedent in the text. The most interesting result of this study from a corpus annotation perspective was the rather low agreement (K=0.63) that we obtained using versions of Hawkins' and Prince's classification schemes; better results (K=0.76) were obtained using the simplified scheme proposed by Fraurud that includes only two classes, first-mention and subsequent-mention. The agreement about antecedents was also not complete. These findings raise questions concerning the strategy of evaluating systems for definite description interpretation by comparing their results with a standardized annotation. From a linguistic point of view, the most interesting observations were the great number of discourse-new definites in our corpus (in one of our experiments, about 50% of the definites in the collection were classified as discourse-new, 30% as anaphoric, and 18% as associative/bridging) and the presence of definites which did not seem to require a complete disambiguation.
cmp-lg/9710008
Probabilistic Event Categorization
cmp-lg cs.CL
This paper describes the automation of a new text categorization task. The categories assigned in this task are more syntactically, semantically, and contextually complex than those typically assigned by fully automatic systems that process unseen test data. Our system for assigning these categories is a probabilistic classifier, developed with a recent method for formulating a probabilistic model from a predefined set of potential features. This paper focuses on feature selection. It presents a number of fully automatic features. It identifies and evaluates various approaches to organizing collocational properties into features, and presents the results of experiments covarying type of organization and type of property. We find that one organization is not best for all kinds of properties, so this is an experimental parameter worth investigating in NLP systems. In addition, the results suggest a way to take advantage of properties that are low frequency but strongly indicative of a class. The problems of recognizing and organizing the various kinds of contextual information required to perform a linguistically complex categorization task have rarely been systematically investigated in NLP.
cmp-lg/9711001
Probabilistic Constraint Logic Programming
cmp-lg cs.CL
This paper addresses two central problems for probabilistic processing models: parameter estimation from incomplete data and efficient retrieval of most probable analyses. These questions have been answered satisfactorily only for probabilistic regular and context-free models. We address these problems for a more expressive probabilistic constraint logic programming model. We present a log-linear probability model for probabilistic constraint logic programming. On top of this model we define an algorithm to estimate the parameters and to select the properties of log-linear models from incomplete data. This algorithm is an extension of the improved iterative scaling algorithm of Della-Pietra, Della-Pietra, and Lafferty (1995). Our algorithm applies to log-linear models in general and is accompanied with suitable approximation methods when applied to large data spaces. Furthermore, we present an approach for searching for most probable analyses of the probabilistic constraint logic programming model. This method can be applied to the ambiguity resolution problem in natural language processing applications.
cmp-lg/9711002
Approximating Context-Free Grammars with a Finite-State Calculus
cmp-lg cs.CL
Although adequate models of human language for syntactic analysis and semantic interpretation are of at least context-free complexity, for applications such as speech processing in which speed is important finite-state models are often preferred. These requirements may be reconciled by using the more complex grammar to automatically derive a finite-state approximation which can then be used as a filter to guide speech recognition or to reject many hypotheses at an early stage of processing. A method is presented here for calculating such finite-state approximations from context-free grammars. It is essentially different from the algorithm introduced by Pereira and Wright (1991; 1996), is faster in some cases, and has the advantage of being open-ended and adaptable.
cmp-lg/9711003
Probabilistic Parsing Using Left Corner Language Models
cmp-lg cs.CL
We introduce a novel parser based on a probabilistic version of a left-corner parser. The left-corner strategy is attractive because rule probabilities can be conditioned on both top-down goals and bottom-up derivations. We develop the underlying theory and explain how a grammar can be induced from analyzed data. We show that the left-corner approach provides an advantage over simple top-down probabilistic context-free grammars in parsing the Wall Street Journal using a grammar induced from the Penn Treebank. We also conclude that the Penn Treebank provides a fairly weak testbed due to the flatness of its bracketings and to the obvious overgeneration and undergeneration of its induced grammar.
cmp-lg/9711004
Variation and Synthetic Speech
cmp-lg cs.CL
We describe the approach to linguistic variation taken by the Motorola speech synthesizer. A pan-dialectal pronunciation dictionary is described, which serves as the training data for a neural network based letter-to-sound converter. Subsequent to dictionary retrieval or letter-to-sound generation, pronunciations are submitted a neural network based postlexical module. The postlexical module has been trained on aligned dictionary pronunciations and hand-labeled narrow phonetic transcriptions. This architecture permits the learning of individual postlexical variation, and can be retrained for each speaker whose voice is being modeled for synthesis. Learning variation in this way can result in greater naturalness for the synthetic speech that is produced by the system.
cmp-lg/9711005
Some apparently disjoint aims and requirements for grammar development environments: the case of natural language generation
cmp-lg cs.CL
Grammar development environments (GDE's) for analysis and for generation have not yet come together. Despite the fact that analysis-oriented GDE's (such as ALEP) may include some possibility of sentence generation, the development techniques and kinds of resources suggested are apparently not those required for practical, large-scale natural language generation work. Indeed, there is no use of `standard' (i.e., analysis-oriented) GDE's in current projects/applications targetting the generation of fluent, coherent texts. This unsatisfactory situation requires some analysis and explanation, which this paper attempts using as an example an extensive GDE for generation. The support provided for distributed large-scale grammar development, multilinguality, and resource maintenance are discussed and contrasted with analysis-oriented approaches.
cmp-lg/9711006
Contextual Information and Specific Language Models for Spoken Language Understanding
cmp-lg cs.CL
In this paper we explain how contextual expectations are generated and used in the task-oriented spoken language understanding system Dialogos. The hard task of recognizing spontaneous speech on the telephone may greatly benefit from the use of specific language models during the recognition of callers' utterances. By 'specific language models' we mean a set of language models that are trained on contextually appropriated data, and that are used during different states of the dialogue on the basis of the information sent to the acoustic level by the dialogue management module. In this paper we describe how the specific language models are obtained on the basis of contextual information. The experimental result we report show that recognition and understanding performance are improved thanks to the use of specific language models.
cmp-lg/9711007
Language Modelling For Task-Oriented Domains
cmp-lg cs.CL
This paper is focused on the language modelling for task-oriented domains and presents an accurate analysis of the utterances acquired by the Dialogos spoken dialogue system. Dialogos allows access to the Italian Railways timetable by using the telephone over the public network. The language modelling aspects of specificity and behaviour to rare events are studied. A technique for getting a language model more robust, based on sentences generated by grammars, is presented. Experimental results show the benefit of the proposed technique. The increment of performance between language models created using grammars and usual ones, is higher when the amount of training material is limited. Therefore this technique can give an advantage especially for the development of language models in a new domain.
cmp-lg/9711008
On the use of expectations for detecting and repairing human-machine miscommunication
cmp-lg cs.CL
In this paper I describe how miscommunication problems are dealt with in the spoken language system DIALOGOS. The dialogue module of the system exploits dialogic expectations in a twofold way: to model what future user utterance might be about (predictions), and to account how the user's next utterance may be related to previous ones in the ongoing interaction (pragmatic-based expectations). The analysis starts from the hypothesis that the occurrence of miscommunication is concomitant with two pragmatic phenomena: the deviation of the user from the expected behaviour and the generation of a conversational implicature. A preliminary evaluation of a large amount of interactions between subjects and DIALOGOS shows that the system performance is enhanced by the uses of both predictions and pragmatic-based expectations.
cmp-lg/9711009
Towards an Improved Performance Measure for Language Models
cmp-lg cs.CL
In this paper a first attempt at deriving an improved performance measure for language models, the probability ratio measure (PRM) is described. In a proof of concept experiment, it is shown that PRM correlates better with recognition accuracy and can lead to better recognition results when used as the optimisation criterion of a clustering algorithm. Inspite of the approximations and limitations of this preliminary work, the results are very encouraging and should justify more work along the same lines.
cmp-lg/9711010
Application-driven automatic subgrammar extraction
cmp-lg cs.CL
The space and run-time requirements of broad coverage grammars appear for many applications unreasonably large in relation to the relative simplicity of the task at hand. On the other hand, handcrafted development of application-dependent grammars is in danger of duplicating work which is then difficult to re-use in other contexts of application. To overcome this problem, we present in this paper a procedure for the automatic extraction of application-tuned consistent subgrammars from proved large-scale generation grammars. The procedure has been implemented for large-scale systemic grammars and builds on the formal equivalence between systemic grammars and typed unification based grammars. Its evaluation for the generation of encyclopedia entries is described, and directions of future development, applicability, and extensions are discussed.
cmp-lg/9711011
The effect of alternative tree representations on tree bank grammars
cmp-lg cs.CL
The performance of PCFGs estimated from tree banks is sensitive to the particular way in which linguistic constructions are represented as trees in the tree bank. This paper presents a theoretical analysis of the effect of different tree representations for PP attachment on PCFG models, and introduces a new methodology for empirically examining such effects using tree transformations. It shows that one transformation, which copies the label of a parent node onto the labels of its children, can improve the performance of a PCFG model in terms of labelled precision and recall on held out data from 73% (precision) and 69% (recall) to 80% and 79% respectively. It also points out that if only maximum likelihood parses are of interest then many productions can be ignored, since they are subsumed by combinations of other productions in the grammar. In the Penn II tree bank grammar, almost 9% of productions are subsumed in this way.
cmp-lg/9711012
Proof Nets and the Complexity of Processing Center-Embedded Constructions
cmp-lg cs.CL
This paper shows how proof nets can be used to formalize the notion of ``incomplete dependency'' used in psycholinguistic theories of the unacceptability of center-embedded constructions. Such theories of human language processing can usually be restated in terms of geometrical constraints on proof nets. The paper ends with a discussion of the relationship between these constraints and incremental semantic interpretation.
cmp-lg/9711013
Features as Resources in R-LFG
cmp-lg cs.CL
This paper introduces a non-unification-based version of LFG called R-LFG (Resource-based Lexical Functional Grammar), which combines elements from both LFG and Linear Logic. The paper argues that a resource sensitive account provides a simpler treatment of many linguistic uses of non-monotonic devices in LFG, such as existential constraints and constraint equations.
cmp-lg/9711014
Type-driven semantic interpretation and feature dependencies in R-LFG
cmp-lg cs.CL
Once one has enriched LFG's formal machinery with the linear logic mechanisms needed for semantic interpretation as proposed by Dalrymple et. al., it is natural to ask whether these make any existing components of LFG redundant. As Dalrymple and her colleagues note, LFG's f-structure completeness and coherence constraints fall out as a by-product of the linear logic machinery they propose for semantic interpretation, thus making those f-structure mechanisms redundant. Given that linear logic machinery or something like it is independently needed for semantic interpretation, it seems reasonable to explore the extent to which it is capable of handling feature structure constraints as well. R-LFG represents the extreme position that all linguistically required feature structure dependencies can be captured by the resource-accounting machinery of a linear or similiar logic independently needed for semantic interpretation, making LFG's unification machinery redundant. The goal is to show that LFG linguistic analyses can be expressed as clearly and perspicuously using the smaller set of mechanisms of R-LFG as they can using the much larger set of unification-based mechanisms in LFG: if this is the case then we will have shown that positing these extra f-structure mechanisms is not linguistically warranted.
cmp-lg/9712001
Applying Explanation-based Learning to Control and Speeding-up Natural Language Generation
cmp-lg cs.CL
This paper presents a method for the automatic extraction of subgrammars to control and speeding-up natural language generation NLG. The method is based on explanation-based learning (EBL). The main advantage for the proposed new method for NLG is that the complexity of the grammatical decision making process during NLG can be vastly reduced, because the EBL method supports the adaption of a NLG system to a particular use of a language.
cmp-lg/9712002
Machine Learning of User Profiles: Representational Issues
cmp-lg cs.CL cs.LG
As more information becomes available electronically, tools for finding information of interest to users becomes increasingly important. The goal of the research described here is to build a system for generating comprehensible user profiles that accurately capture user interest with minimum user interaction. The research described here focuses on the importance of a suitable generalization hierarchy and representation for learning profiles which are predictively accurate and comprehensible. In our experiments we evaluated both traditional features based on weighted term vectors as well as subject features corresponding to categories which could be drawn from a thesaurus. Our experiments, conducted in the context of a content-based profiling system for on-line newspapers on the World Wide Web (the IDD News Browser), demonstrate the importance of a generalization hierarchy and the promise of combining natural language processing techniques with machine learning (ML) to address an information retrieval (IR) problem.
cmp-lg/9712003
Context as a Spurious Concept
cmp-lg cs.CL
I take issue with AI formalizations of context, primarily the formalization by McCarthy and Buvac, that regard context as an undefined primitive whose formalization can be the same in many different kinds of AI tasks. In particular, any theory of context in natural language must take the special nature of natural language into account and cannot regard context simply as an undefined primitive. I show that there is no such thing as a coherent theory of context simpliciter -- context pure and simple -- and that context in natural language is not the same kind of thing as context in KR. In natural language, context is constructed by the speaker and the interpreter, and both have considerable discretion in so doing. Therefore, a formalization based on pre-defined contexts and pre-defined `lifting axioms' cannot account for how context is used in real-world language.
cmp-lg/9712004
Multi-document Summarization by Graph Search and Matching
cmp-lg cs.CL
We describe a new method for summarizing similarities and differences in a pair of related documents using a graph representation for text. Concepts denoted by words, phrases, and proper names in the document are represented positionally as nodes in the graph along with edges corresponding to semantic relations between items. Given a perspective in terms of which the pair of documents is to be summarized, the algorithm first uses a spreading activation technique to discover, in each document, nodes semantically related to the topic. The activated graphs of each document are then matched to yield a graph corresponding to similarities and differences between the pair, which is rendered in natural language. An evaluation of these techniques has been carried out.
cmp-lg/9712005
Topic Graph Generation for Query Navigation: Use of Frequency Classes for Topic Extraction
cmp-lg cs.CL
To make an interactive guidance mechanism for document retrieval systems, we developed a user-interface which presents users the visualized map of topics at each stage of retrieval process. Topic words are automatically extracted by frequency analysis and the strength of the relationships between topic words is measured by their co-occurrence. A major factor affecting a user's impression of a given topic word graph is the balance between common topic words and specific topic words. By using frequency classes for topic word extraction, we made it possible to select well-balanced set of topic words, and to adjust the balance of common and specific topic words.
cmp-lg/9712006
"I don't believe in word senses"
cmp-lg cs.CL
Word sense disambiguation assumes word senses. Within the lexicography and linguistics literature, they are known to be very slippery entities. The paper looks at problems with existing accounts of `word sense' and describes the various kinds of ways in which a word's meaning can deviate from its core meaning. An analysis is presented in which word senses are abstractions from clusters of corpus citations, in accordance with current lexicographic practice. The corpus citations, not the word senses, are the basic objects in the ontology. The corpus citations will be clustered into senses according to the purposes of whoever or whatever does the clustering. In the absence of such purposes, word senses do not exist. Word sense disambiguation also needs a set of word senses to disambiguate between. In most recent work, the set has been taken from a general-purpose lexical resource, with the assumption that the lexical resource describes the word senses of English/French/..., between which NLP applications will need to disambiguate. The implication of the paper is, by contrast, that word senses exist only relative to a task.
cmp-lg/9712007
Foreground and Background Lexicons and Word Sense Disambiguation for Information Extraction
cmp-lg cs.CL
Lexicon acquisition from machine-readable dictionaries and corpora is currently a dynamic field of research, yet it is often not clear how lexical information so acquired can be used, or how it relates to structured meaning representations. In this paper I look at this issue in relation to Information Extraction (hereafter IE), and one subtask for which both lexical and general knowledge are required, Word Sense Disambiguation (WSD). The analysis is based on the widely-used, but little-discussed distinction between an IE system's foreground lexicon, containing the domain's key terms which map onto the database fields of the output formalism, and the background lexicon, containing the remainder of the vocabulary. For the foreground lexicon, human lexicography is required. For the background lexicon, automatic acquisition is appropriate. For the foreground lexicon, WSD will occur as a by-product of finding a coherent semantic interpretation of the input. WSD techniques as discussed in recent literature are suited only to the background lexicon. Once the foreground/background distinction is developed, there is a match between what is possible, given the state of the art in WSD, and what is required, for high-quality IE.
cmp-lg/9712008
What is word sense disambiguation good for?
cmp-lg cs.CL
Word sense disambiguation has developed as a sub-area of natural language processing, as if, like parsing, it was a well-defined task which was a pre-requisite to a wide range of language-understanding applications. First, I review earlier work which shows that a set of senses for a word is only ever defined relative to a particular human purpose, and that a view of word senses as part of the linguistic furniture lacks theoretical underpinnings. Then, I investigate whether and how word sense ambiguity is in fact a problem for different varieties of NLP application.
cmp-lg/9712009
Speech Repairs, Intonational Boundaries and Discourse Markers: Modeling Speakers' Utterances in Spoken Dialog
cmp-lg cs.CL
In this thesis, we present a statistical language model for resolving speech repairs, intonational boundaries and discourse markers. Rather than finding the best word interpretation for an acoustic signal, we redefine the speech recognition problem to so that it also identifies the POS tags, discourse markers, speech repairs and intonational phrase endings (a major cue in determining utterance units). Adding these extra elements to the speech recognition problem actually allows it to better predict the words involved, since we are able to make use of the predictions of boundary tones, discourse markers and speech repairs to better account for what word will occur next. Furthermore, we can take advantage of acoustic information, such as silence information, which tends to co-occur with speech repairs and intonational phrase endings, that current language models can only regard as noise in the acoustic signal. The output of this language model is a much fuller account of the speaker's turn, with part-of-speech assigned to each word, intonation phrase endings and discourse markers identified, and speech repairs detected and corrected. In fact, the identification of the intonational phrase endings, discourse markers, and resolution of the speech repairs allows the speech recognizer to model the speaker's utterances, rather than simply the words involved, and thus it can return a more meaningful analysis of the speaker's turn for later processing.
cmp-lg/9712010
Orthographic Structuring of Human Speech and Texts: Linguistic Application of Recurrence Quantification Analysis
cmp-lg cs.CL
A methodology based upon recurrence quantification analysis is proposed for the study of orthographic structure of written texts. Five different orthographic data sets (20th century Italian poems, 20th century American poems, contemporary Swedish poems with their corresponding Italian translations, Italian speech samples, and American speech samples) were subjected to recurrence quantification analysis, a procedure which has been found to be diagnostically useful in the quantitative assessment of ordered series in fields such as physics, molecular dynamics, physiology, and general signal processing. Recurrence quantification was developed from recurrence plots as applied to the analysis of nonlinear, complex systems in the physical sciences, and is based on the computation of a distance matrix of the elements of an ordered series (in this case the letters consituting selected speech and poetic texts). From a strictly mathematical view, the results show the possibility of demonstrating invariance between different language exemplars despite the apparent low-level of coding (orthography). Comparison with the actual texts confirms the ability of the method to reveal recurrent structures, and their complexity. Using poems as a reference standard for judging speech complexity, the technique exhibits language independence, order dependence and freedom from pure statistical characteristics of studied sequences, as well as consistency with easily identifiable texts. Such studies may provide phenomenological markers of hidden structure as coded by the purely orthographic level.
cmp-lg/9801001
Hierarchical Non-Emitting Markov Models
cmp-lg cs.CL
We describe a simple variant of the interpolated Markov model with non-emitting state transitions and prove that it is strictly more powerful than any Markov model. More importantly, the non-emitting model outperforms the classic interpolated model on the natural language texts under a wide range of experimental conditions, with only a modest increase in computational requirements. The non-emitting model is also much less prone to overfitting. Keywords: Markov model, interpolated Markov model, hidden Markov model, mixture modeling, non-emitting state transitions, state-conditional interpolation, statistical language model, discrete time series, Brown corpus, Wall Street Journal.
cmp-lg/9801002
Identifying Discourse Markers in Spoken Dialog
cmp-lg cs.CL
In this paper, we present a method for identifying discourse marker usage in spontaneous speech based on machine learning. Discourse markers are denoted by special POS tags, and thus the process of POS tagging can be used to identify discourse markers. By incorporating POS tagging into language modeling, discourse markers can be identified during speech recognition, in which the timeliness of the information can be used to help predict the following words. We contrast this approach with an alternative machine learning approach proposed by Litman (1996). This paper also argues that discourse markers can be used to help the hearer predict the role that the upcoming utterance plays in the dialog. Thus discourse markers should provide valuable evidence for automatic dialog act prediction.
cmp-lg/9801003
Do not forget: Full memory in memory-based learning of word pronunciation
cmp-lg cs.CL
Memory-based learning, keeping full memory of learning material, appears a viable approach to learning NLP tasks, and is often superior in generalisation accuracy to eager learning approaches that abstract from learning material. Here we investigate three partial memory-based learning approaches which remove from memory specific task instance types estimated to be exceptional. The three approaches each implement one heuristic function for estimating exceptionality of instance types: (i) typicality, (ii) class prediction strength, and (iii) friendly-neighbourhood size. Experiments are performed with the memory-based learning algorithm IB1-IG trained on English word pronunciation. We find that removing instance types with low prediction strength (ii) is the only tested method which does not seriously harm generalisation accuracy. We conclude that keeping full memory of types rather than tokens, and excluding minority ambiguities appear to be the only performance-preserving optimisations of memory-based learning.
cmp-lg/9801004
Modularity in inductively-learned word pronunciation systems
cmp-lg cs.CL
In leading morpho-phonological theories and state-of-the-art text-to-speech systems it is assumed that word pronunciation cannot be learned or performed without in-between analyses at several abstraction levels (e.g., morphological, graphemic, phonemic, syllabic, and stress levels). We challenge this assumption for the case of English word pronunciation. Using IGTree, an inductive-learning decision-tree algorithms, we train and test three word-pronunciation systems in which the number of abstraction levels (implemented as sequenced modules) is reduced from five, via three, to one. The latter system, classifying letter strings directly as mapping to phonemes with stress markers, yields significantly better generalisation accuracies than the two multi-module systems. Analyses of empirical results indicate that positive utility effects of sequencing modules are outweighed by cascading errors passed on between modules.
cmp-lg/9801005
A General, Sound and Efficient Natural Language Parsing Algorithm based on Syntactic Constraints Propagation
cmp-lg cs.CL
This paper presents a new context-free parsing algorithm based on a bidirectional strictly horizontal strategy which incorporates strong top-down predictions (derivations and adjacencies). From a functional point of view, the parser is able to propagate syntactic constraints reducing parsing ambiguity. From a computational perspective, the algorithm includes different techniques aimed at the improvement of the manipulation and representation of the structures used.
cmp-lg/9802001
Look-Back and Look-Ahead in the Conversion of Hidden Markov Models into Finite State Transducers
cmp-lg cs.CL
This paper describes the conversion of a Hidden Markov Model into a finite state transducer that closely approximates the behavior of the stochastic model. In some cases the transducer is equivalent to the HMM. This conversion is especially advantageous for part-of-speech tagging because the resulting transducer can be composed with other transducers that encode correction rules for the most frequent tagging errors. The speed of tagging is also improved. The described methods have been implemented and successfully tested.
cmp-lg/9802002
A Hybrid Environment for Syntax-Semantic Tagging
cmp-lg cs.CL
The thesis describes the application of the relaxation labelling algorithm to NLP disambiguation. Language is modelled through context constraint inspired on Constraint Grammars. The constraints enable the use of a real value statind "compatibility". The technique is applied to POS tagging, Shallow Parsing and Word Sense Disambigation. Experiments and results are reported. The proposed approach enables the use of multi-feature constraint models, the simultaneous resolution of several NL disambiguation tasks, and the collaboration of linguistic and statistical models.
cmp-lg/9803001
Automating Coreference: The Role of Annotated Training Data
cmp-lg cs.CL
We report here on a study of interannotator agreement in the coreference task as defined by the Message Understanding Conference (MUC-6 and MUC-7). Based on feedback from annotators, we clarified and simplified the annotation specification. We then performed an analysis of disagreement among several annotators, concluding that only 16% of the disagreements represented genuine disagreement about coreference; the remainder of the cases were mostly typographical errors or omissions, easily reconciled. Initially, we measured interannotator agreement in the low 80s for precision and recall. To try to improve upon this, we ran several experiments. In our final experiment, we separated the tagging of candidate noun phrases from the linking of actual coreferring expressions. This method shows promise - interannotator agreement climbed to the low 90s - but it needs more extensive validation. These results position the research community to broaden the coreference task to multiple languages, and possibly to different kinds of coreference.
cmp-lg/9803002
Time, Tense and Aspect in Natural Language Database Interfaces
cmp-lg cs.CL
Most existing natural language database interfaces (NLDBs) were designed to be used with database systems that provide very limited facilities for manipulating time-dependent data, and they do not support adequately temporal linguistic mechanisms (verb tenses, temporal adverbials, temporal subordinate clauses, etc.). The database community is becoming increasingly interested in temporal database systems, that are intended to store and manipulate in a principled manner information not only about the present, but also about the past and future. When interfacing to temporal databases, supporting temporal linguistic mechanisms becomes crucial. We present a framework for constructing natural language interfaces for temporal databases (NLTDBs), that draws on research in tense and aspect theories, temporal logics, and temporal databases. The framework consists of a temporal intermediate representation language, called TOP, an HPSG grammar that maps a wide range of questions involving temporal mechanisms to appropriate TOP expressions, and a provably correct method for translating from TOP to TSQL2, TSQL2 being a recently proposed temporal extension of the SQL database language. This framework was employed to implement a prototype NLTDB using ALE and Prolog.
cmp-lg/9803003
Nymble: a High-Performance Learning Name-finder
cmp-lg cs.CL
This paper presents a statistical, learned approach to finding names and other non-recursive entities in text (as per the MUC-6 definition of the NE task), using a variant of the standard hidden Markov model. We present our justification for the problem and our approach, a detailed discussion of the model itself and finally the successful results of this new approach.
cmp-lg/9804001
Graph Interpolation Grammars: a Rule-based Approach to the Incremental Parsing of Natural Languages
cmp-lg cs.CL
Graph Interpolation Grammars are a declarative formalism with an operational semantics. Their goal is to emulate salient features of the human parser, and notably incrementality. The parsing process defined by GIGs incrementally builds a syntactic representation of a sentence as each successive lexeme is read. A GIG rule specifies a set of parse configurations that trigger its application and an operation to perform on a matching configuration. Rules are partly context-sensitive; furthermore, they are reversible, meaning that their operations can be undone, which allows the parsing process to be nondeterministic. These two factors confer enough expressive power to the formalism for parsing natural languages.
cmp-lg/9804002
The Proper Treatment of Optimality in Computational Phonology
cmp-lg cs.CL
This paper presents a novel formalization of optimality theory. Unlike previous treatments of optimality in computational linguistics, starting with Ellison (1994), the new approach does not require any explicit marking and counting of constraint violations. It is based on the notion of "lenient composition," defined as the combination of ordinary composition and priority union. If an underlying form has outputs that can meet a given constraint, lenient composition enforces the constraint; if none of the output candidates meet the constraint, lenient composition allows all of them. For the sake of greater efficiency, we may "leniently compose" the GEN relation and all the constraints into a single finite-state transducer that maps each underlying form directly into its optimal surface realizations, and vice versa, without ever producing any failing candidates. Seen from this perspective, optimality theory is surprisingly similar to the two older strains of finite-state phonology: classical rewrite systems and two-level models. In particular, the ranking of optimality constraints corresponds to the ordering of rewrite rules.
cmp-lg/9804003
Treatment of Epsilon-Moves in Subset Construction
cmp-lg cs.CL
The paper discusses the problem of determinising finite-state automata containing large numbers of epsilon-moves. Experiments with finite-state approximations of natural language grammars often give rise to very large automata with a very large number of epsilon-moves. The paper identifies three subset construction algorithms which treat epsilon-moves. A number of experiments has been performed which indicate that the algorithms differ considerably in practice. Furthermore, the experiments suggest that the average number of epsilon-moves per state can be used to predict which algorithm is likely to perform best for a given input automaton.
cmp-lg/9804004
Corpus-Based Word Sense Disambiguation
cmp-lg cs.CL
Resolution of lexical ambiguity, commonly termed ``word sense disambiguation'', is expected to improve the analytical accuracy for tasks which are sensitive to lexical semantics. Such tasks include machine translation, information retrieval, parsing, natural language understanding and lexicography. Reflecting the growth in utilization of machine readable texts, word sense disambiguation techniques have been explored variously in the context of corpus-based approaches. Within one corpus-based framework, that is the similarity-based method, systems use a database, in which example sentences are manually annotated with correct word senses. Given an input, systems search the database for the most similar example to the input. The lexical ambiguity of a word contained in the input is resolved by selecting the sense annotation of the retrieved example. In this research, we apply this method of resolution of verbal polysemy, in which the similarity between two examples is computed as the weighted average of the similarity between complements governed by a target polysemous verb. We explore similarity-based verb sense disambiguation focusing on the following three methods. First, we propose a weighting schema for each verb complement in the similarity computation. Second, in similarity-based techniques, the overhead for manual supervision and searching the large-sized database can be prohibitive. To resolve this problem, we propose a method to select a small number of effective examples, for system usage. Finally, the efficiency of our system is highly dependent on the similarity computation used. To maximize efficiency, we propose a method which integrates the advantages of previous methods for similarity computation.
cmp-lg/9804005
On the existence of certain total recursive functions in nontrivial axiom systems, I
cmp-lg cs.CL
We investigate the existence of a class of ZFC-provably total recursive unary functions, given certain constraints, and apply some of those results to show that, for $\Sigma_1$-sound set theory, ZFC$\not\vdash P<NP$.
cmp-lg/9805001
Valence Induction with a Head-Lexicalized PCFG
cmp-lg cs.CL
This paper presents an experiment in learning valences (subcategorization frames) from a 50 million word text corpus, based on a lexicalized probabilistic context free grammar. Distributions are estimated using a modified EM algorithm. We evaluate the acquired lexicon both by comparison with a dictionary and by entropy measures. Results show that our model produces highly accurate frame distributions.
cmp-lg/9805002
Group Theory and Grammatical Description
cmp-lg cs.CL
This paper presents a model for linguistic description based on group theory. A grammar in this model, or "G-grammar", is a collection of lexical expressions which are products of logical forms, phonological forms, and their inverses. Phrasal descriptions are obtained by forming products of lexical expressions and by cancelling contiguous elements which are inverses of each other. We show applications of this model to parsing and generation, long-distance movement, and quantifier scoping. We believe that by moving from the free monoid over a vocabulary V --- standard in formal language studies --- to the free group over V, deep affinities between linguistic phenomena and classical algebra come to the surface, and that the consequences of tapping the mathematical connections thus established could be considerable.
cmp-lg/9805003
Models of Co-occurrence
cmp-lg cs.CL
A model of co-occurrence in bitext is a boolean predicate that indicates whether a given pair of word tokens co-occur in corresponding regions of the bitext space. Co-occurrence is a precondition for the possibility that two tokens might be mutual translations. Models of co-occurrence are the glue that binds methods for mapping bitext correspondence with methods for estimating translation models into an integrated system for exploiting parallel texts. Different models of co-occurrence are possible, depending on the kind of bitext map that is available, the language-specific information that is available, and the assumptions made about the nature of translational equivalence. Although most statistical translation models are based on models of co-occurrence, modeling co-occurrence correctly is more difficult than it may at first appear.
cmp-lg/9805004
Annotation Style Guide for the Blinker Project
cmp-lg cs.CL
This annotation style guide was created by and for the Blinker project at the University of Pennsylvania. The Blinker project was so named after the ``bilingual linker'' GUI, which was created to enable bilingual annotators to ``link'' word tokens that are mutual translations in parallel texts. The parallel text chosen for this project was the Bible, because it is probably the easiest text to obtain in electronic form in multiple languages. The languages involved were English and French, because, of the languages with which the project co-ordinator was familiar, these were the two for which a sufficient number of annotators was likely to be found.
cmp-lg/9805005
Manual Annotation of Translational Equivalence: The Blinker Project
cmp-lg cs.CL
Bilingual annotators were paid to link roughly sixteen thousand corresponding words between on-line versions of the Bible in modern French and modern English. These annotations are freely available to the research community from http://www.cis.upenn.edu/~melamed . The annotations can be used for several purposes. First, they can be used as a standard data set for developing and testing translation lexicons and statistical translation models. Second, researchers in lexical semantics will be able to mine the annotations for insights about cross-linguistic lexicalization patterns. Third, the annotations can be used in research into certain recently proposed methods for monolingual word-sense disambiguation. This paper describes the annotated texts, the specially-designed annotation tool, and the strategies employed to increase the consistency of the annotations. The annotation process was repeated five times by different annotators. Inter-annotator agreement rates indicate that the annotations are reasonably reliable and that the method is easy to replicate.
cmp-lg/9805006
Word-to-Word Models of Translational Equivalence
cmp-lg cs.CL
Parallel texts (bitexts) have properties that distinguish them from other kinds of parallel data. First, most words translate to only one other word. Second, bitext correspondence is noisy. This article presents methods for biasing statistical translation models to reflect these properties. Analysis of the expected behavior of these biases in the presence of sparse data predicts that they will result in more accurate models. The prediction is confirmed by evaluation with respect to a gold standard -- translation models that are biased in this fashion are significantly more accurate than a baseline knowledge-poor model. This article also shows how a statistical translation model can take advantage of various kinds of pre-existing knowledge that might be available about particular language pairs. Even the simplest kinds of language-specific knowledge, such as the distinction between content words and function words, is shown to reliably boost translation model performance on some tasks. Statistical models that are informed by pre-existing knowledge about the model domain combine the best of both the rationalist and empiricist traditions.
cmp-lg/9805007
Parsing Inside-Out
cmp-lg cs.CL
The inside-outside probabilities are typically used for reestimating Probabilistic Context Free Grammars (PCFGs), just as the forward-backward probabilities are typically used for reestimating HMMs. I show several novel uses, including improving parser accuracy by matching parsing algorithms to evaluation criteria; speeding up DOP parsing by 500 times; and 30 times faster PCFG thresholding at a given accuracy level. I also give an elegant, state-of-the-art grammar formalism, which can be used to compute inside-outside probabilities; and a parser description formalism, which makes it easy to derive inside-outside formulas and many others.
cmp-lg/9805008
A Descriptive Characterization of Tree-Adjoining Languages (Full Version)
cmp-lg cs.CL
Since the early Sixties and Seventies it has been known that the regular and context-free languages are characterized by definability in the monadic second-order theory of certain structures. More recently, these descriptive characterizations have been used to obtain complexity results for constraint- and principle-based theories of syntax and to provide a uniform model-theoretic framework for exploring the relationship between theories expressed in disparate formal terms. These results have been limited, to an extent, by the lack of descriptive characterizations of language classes beyond the context-free. Recently, we have shown that tree-adjoining languages (in a mildly generalized form) can be characterized by recognition by automata operating on three-dimensional tree manifolds, a three-dimensional analog of trees. In this paper, we exploit these automata-theoretic results to obtain a characterization of the tree-adjoining languages by definability in the monadic second-order theory of these three-dimensional tree manifolds. This not only opens the way to extending the tools of model-theoretic syntax to the level of TALs, but provides a highly flexible mechanism for defining TAGs in terms of logical constraints. This is the full version of a paper to appear in the proceedings of COLING-ACL'98 as a project note.
cmp-lg/9805009
Discovery of Linguistic Relations Using Lexical Attraction
cmp-lg cs.CL
This work has been motivated by two long term goals: to understand how humans learn language and to build programs that can understand language. Using a representation that makes the relevant features explicit is a prerequisite for successful learning and understanding. Therefore, I chose to represent relations between individual words explicitly in my model. Lexical attraction is defined as the likelihood of such relations. I introduce a new class of probabilistic language models named lexical attraction models which can represent long distance relations between words and I formalize this new class of models using information theory. Within the framework of lexical attraction, I developed an unsupervised language acquisition program that learns to identify linguistic relations in a given sentence. The only explicitly represented linguistic knowledge in the program is lexical attraction. There is no initial grammar or lexicon built in and the only input is raw text. Learning and processing are interdigitated. The processor uses the regularities detected by the learner to impose structure on the input. This structure enables the learner to detect higher level regularities. Using this bootstrapping procedure, the program was trained on 100 million words of Associated Press material and was able to achieve 60% precision and 50% recall in finding relations between content-words. Using knowledge of lexical attraction, the program can identify the correct relations in syntactically ambiguous sentences such as ``I saw the Statue of Liberty flying over New York.''
cmp-lg/9805010
Integrating Text Plans for Conciseness and Coherence
cmp-lg cs.CL
Our experience with a critiquing system shows that when the system detects problems with the user's performance, multiple critiques are often produced. Analysis of a corpus of actual critiques revealed that even though each individual critique is concise and coherent, the set of critiques as a whole may exhibit several problems that detract from conciseness and coherence, and consequently assimilation. Thus a text planner was needed that could integrate the text plans for individual communicative goals to produce an overall text plan representing a concise, coherent message. This paper presents our general rule-based system for accomplishing this task. The system takes as input a \emph{set} of individual text plans represented as RST-style trees, and produces a smaller set of more complex trees representing integrated messages that still achieve the multiple communicative goals of the individual text plans. Domain-independent rules are used to capture strategies across domains, while the facility for addition of domain-dependent rules enables the system to be tuned to the requirements of a particular domain. The system has been tested on a corpus of critiques in the domain of trauma care.
cmp-lg/9805011
Automatic summarising: factors and directions
cmp-lg cs.CL
This position paper suggests that progress with automatic summarising demands a better research methodology and a carefully focussed research strategy. In order to develop effective procedures it is necessary to identify and respond to the context factors, i.e. input, purpose, and output factors, that bear on summarising and its evaluation. The paper analyses and illustrates these factors and their implications for evaluation. It then argues that this analysis, together with the state of the art and the intrinsic difficulty of summarising, imply a nearer-term strategy concentrating on shallow, but not surface, text analysis and on indicative summarising. This is illustrated with current work, from which a potentially productive research programme can be developed.
cmp-lg/9805012
Recognizing Syntactic Errors in the Writing of Second Language Learners
cmp-lg cs.CL
This paper reports on the recognition component of an intelligent tutoring system that is designed to help foreign language speakers learn standard English. The system models the grammar of the learner, with this instantiation of the system tailored to signers of American Sign Language (ASL). We discuss the theoretical motivations for the system, various difficulties that have been encountered in the implementation, as well as the methods we have used to overcome these problems. Our method of capturing ungrammaticalities involves using mal-rules (also called 'error productions'). However, the straightforward addition of some mal-rules causes significant performance problems with the parser. For instance, the ASL population has a strong tendency to drop pronouns and the auxiliary verb `to be'. Being able to account for these as sentences results in an explosion in the number of possible parses for each sentence. This explosion, left unchecked, greatly hampers the performance of the system. We discuss how this is handled by taking into account expectations from the specific population (some of which are captured in our unique user model). The different representations of lexical items at various points in the acquisition process are modeled by using mal-rules, which obviates the need for multiple lexicons. The grammar is evaluated on its ability to correctly diagnose agreement problems in actual sentences produced by ASL native speakers.
cmp-lg/9806001
Learning Correlations between Linguistic Indicators and Semantic Constraints: Reuse of Context-Dependent Descriptions of Entities
cmp-lg cs.CL
This paper presents the results of a study on the semantic constraints imposed on lexical choice by certain contextual indicators. We show how such indicators are computed and how correlations between them and the choice of a noun phrase description of a named entity can be automatically established using supervised learning. Based on this correlation, we have developed a technique for automatic lexical choice of descriptions of entities in text generation. We discuss the underlying relationship between the pragmatics of choosing an appropriate description that serves a specific purpose in the automatically generated text and the semantics of the description itself. We present our work in the framework of the more general concept of reuse of linguistic structures that are automatically extracted from large corpora. We present a formal evaluation of our approach and we conclude with some thoughts on potential applications of our method.
cmp-lg/9806002
Computing Dialogue Acts from Features with Transformation-Based Learning
cmp-lg cs.CL
To interpret natural language at the discourse level, it is very useful to accurately recognize dialogue acts, such as SUGGEST, in identifying speaker intentions. Our research explores the utility of a machine learning method called Transformation-Based Learning (TBL) in computing dialogue acts, because TBL has a number of advantages over alternative approaches for this application. We have identified some extensions to TBL that are necessary in order to address the limitations of the original algorithm and the particular demands of discourse processing. We use a Monte Carlo strategy to increase the applicability of the TBL method, and we select features of utterances that can be used as input to improve the performance of TBL. Our system is currently being tested on the VerbMobil corpora of spoken dialogues, producing promising preliminary results.
cmp-lg/9806003
Lazy Transformation-Based Learning
cmp-lg cs.CL
We introduce a significant improvement for a relatively new machine learning method called Transformation-Based Learning. By applying a Monte Carlo strategy to randomly sample from the space of rules, rather than exhaustively analyzing all possible rules, we drastically reduce the memory and time costs of the algorithm, without compromising accuracy on unseen data. This enables Transformation- Based Learning to apply to a wider range of domains, as it can effectively consider a larger number of different features and feature interactions in the data. In addition, the Monte Carlo improvement decreases the labor demands on the human developer, who no longer needs to develop a minimal set of rule templates to maintain tractability.
cmp-lg/9806004
Rationality, Cooperation and Conversational Implicature
cmp-lg cs.CL
Conversational implicatures are usually described as being licensed by the disobeying or flouting of a Principle of Cooperation. However, the specification of this principle has proved computationally elusive. In this paper we suggest that a more useful concept is rationality. Such a concept can be specified explicitely in planning terms and we argue that speakers perform utterances as part of the optimal plan for their particular communicative goals. Such an assumption can be used by the hearer to infer conversational implicatures implicit in the speaker's utterance.
cmp-lg/9806005
Eliminating deceptions and mistaken belief to infer conversational implicature
cmp-lg cs.CL
Conversational implicatures are usually described as being licensed by the disobeying or flouting of some principle by the speaker in cooperative dialogue. However, such work has failed to distinguish cases of the speaker flouting such a principle from cases where the speaker is either deceptive or holds a mistaken belief. In this paper, we demonstrate how the three different cases can be distinguished in terms of the beliefs ascribed to the speaker of the utterance. We argue that in the act of distinguishing the speaker's intention and ascribing such beliefs, the intended inference can be made by the hearer. This theory is implemented in ViewGen, a pre-existing belief modelling system used in a medical counselling domain.
cmp-lg/9806006
Dialogue Act Tagging with Transformation-Based Learning
cmp-lg cs.CL
For the task of recognizing dialogue acts, we are applying the Transformation-Based Learning (TBL) machine learning algorithm. To circumvent a sparse data problem, we extract values of well-motivated features of utterances, such as speaker direction, punctuation marks, and a new feature, called dialogue act cues, which we find to be more effective than cue phrases and word n-grams in practice. We present strategies for constructing a set of dialogue act cues automatically by minimizing the entropy of the distribution of dialogue acts in a training corpus, filtering out irrelevant dialogue act cues, and clustering semantically-related words. In addition, to address limitations of TBL, we introduce a Monte Carlo strategy for training efficiently and a committee method for computing confidence measures. These ideas are combined in our working implementation, which labels held-out data as accurately as any other reported system for the dialogue act tagging task.
cmp-lg/9806007
An Investigation of Transformation-Based Learning in Discourse
cmp-lg cs.CL
This paper presents results from the first attempt to apply Transformation-Based Learning to a discourse-level Natural Language Processing task. To address two limitations of the standard algorithm, we developed a Monte Carlo version of Transformation-Based Learning to make the method tractable for a wider range of problems without degradation in accuracy, and we devised a committee method for assigning confidence measures to tags produced by Transformation-Based Learning. The paper describes these advances, presents experimental evidence that Transformation-Based Learning is as effective as alternative approaches (such as Decision Trees and N-Grams) for a discourse task called Dialogue Act Tagging, and argues that Transformation-Based Learning has desirable features that make it particularly appealing for the Dialogue Act Tagging task.
cmp-lg/9806008
Unlimited Vocabulary Grapheme to Phoneme Conversion for Korean TTS
cmp-lg cs.CL
This paper describes a grapheme-to-phoneme conversion method using phoneme connectivity and CCV conversion rules. The method consists of mainly four modules including morpheme normalization, phrase-break detection, morpheme to phoneme conversion and phoneme connectivity check. The morpheme normalization is to replace non-Korean symbols into standard Korean graphemes. The phrase-break detector assigns phrase breaks using part-of-speech (POS) information. In the morpheme-to-phoneme conversion module, each morpheme in the phrase is converted into phonetic patterns by looking up the morpheme phonetic pattern dictionary which contains candidate phonological changes in boundaries of the morphemes. Graphemes within a morpheme are grouped into CCV patterns and converted into phonemes by the CCV conversion rules. The phoneme connectivity table supports grammaticality checking of the adjacent two phonetic morphemes. In the experiments with a corpus of 4,973 sentences, we achieved 99.9% of the grapheme-to-phoneme conversion performance and 97.5% of the sentence conversion performance. The full Korean TTS system is now being implemented using this conversion method.
cmp-lg/9806009
Methods and Tools for Building the Catalan WordNet
cmp-lg cs.CL
In this paper we introduce the methodology used and the basic phases we followed to develop the Catalan WordNet, and shich lexical resources have been employed in its building. This methodology, as well as the tools we made use of, have been thought in a general way so that they could be applied to any other language.
cmp-lg/9806010
Towards a single proposal is spelling correction
cmp-lg cs.CL
The study presented here relies on the integrated use of different kinds of knowledge in order to improve first-guess accuracy in non-word context-sensitive correction for general unrestricted texts. State of the art spelling correction systems, e.g. ispell, apart from detecting spelling errors, also assist the user by offering a set of candidate corrections that are close to the misspelled word. Based on the correction proposals of ispell, we built several guessers, which were combined in different ways. Firstly, we evaluated all possibilities and selected the best ones in a corpus with artificially generated typing errors. Secondly, the best combinations were tested on texts with genuine spelling errors. The results for the latter suggest that we can expect automatic non-word correction for all the errors in a free running text with 80% precision and a single proposal 98% of the times (1.02 proposals on average).
cmp-lg/9806011
A Memory-Based Approach to Learning Shallow Natural Language Patterns
cmp-lg cs.CL
Recognizing shallow linguistic patterns, such as basic syntactic relationships between words, is a common task in applied natural language and text processing. The common practice for approaching this task is by tedious manual definition of possible pattern structures, often in the form of regular expressions or finite automata. This paper presents a novel memory-based learning method that recognizes shallow patterns in new text based on a bracketed training corpus. The training data are stored as-is, in efficient suffix-tree data structures. Generalization is performed on-line at recognition time by comparing subsequences of the new text to positive and negative evidence in the corpus. This way, no information in the training is lost, as can happen in other learning systems that construct a single generalized model at the time of training. The paper presents experimental results for recognizing noun phrase, subject-verb and verb-object patterns in English. Since the learning approach enables easy porting to new domains, we plan to apply it to syntactic patterns in other languages and to sub-language patterns for information extraction.
cmp-lg/9806012
Bayesian Stratified Sampling to Assess Corpus Utility
cmp-lg cs.CL
This paper describes a method for asking statistical questions about a large text corpus. We exemplify the method by addressing the question, "What percentage of Federal Register documents are real documents, of possible interest to a text researcher or analyst?" We estimate an answer to this question by evaluating 200 documents selected from a corpus of 45,820 Federal Register documents. Stratified sampling is used to reduce the sampling uncertainty of the estimate from over 3100 documents to fewer than 1000. The stratification is based on observed characteristics of real documents, while the sampling procedure incorporates a Bayesian version of Neyman allocation. A possible application of the method is to establish baseline statistics used to estimate recall rates for information retrieval systems.
cmp-lg/9806013
Can Subcategorisation Probabilities Help a Statistical Parser?
cmp-lg cs.CL
Research into the automatic acquisition of lexical information from corpora is starting to produce large-scale computational lexicons containing data on the relative frequencies of subcategorisation alternatives for individual verbal predicates. However, the empirical question of whether this type of frequency information can in practice improve the accuracy of a statistical parser has not yet been answered. In this paper we describe an experiment with a wide-coverage statistical grammar and parser for English and subcategorisation frequencies acquired from ten million words of text which shows that this information can significantly improve parse accuracy.
cmp-lg/9806014
Word Sense Disambiguation using Optimised Combinations of Knowledge Sources
cmp-lg cs.CL
Word sense disambiguation algorithms, with few exceptions, have made use of only one lexical knowledge source. We describe a system which performs unrestricted word sense disambiguation (on all content words in free text) by combining different knowledge sources: semantic preferences, dictionary definitions and subject/domain codes along with part-of-speech tags. The usefulness of these sources is optimised by means of a learning algorithm. We also describe the creation of a new sense tagged corpus by combining existing resources. Tested accuracy of our approach on this corpus exceeds 92%, demonstrating the viability of all-word disambiguation rather than restricting oneself to a small sample.
cmp-lg/9806015
Building Accurate Semantic Taxonomies from Monolingual MRDs
cmp-lg cs.CL
This paper presents a method that combines a set of unsupervised algorithms in order to accurately build large taxonomies from any machine-readable dictionary (MRD). Our aim is to profit from conventional MRDs, with no explicit semantic coding. We propose a system that 1) performs fully automatic exraction of taxonomic links from MRD entries and 2) ranks the extracted relations in a way that selective manual refinement is allowed. Tested accuracy can reach around 100% depending on the degree of coverage selected, showing that taxonomy building is not limited to structured dictionaries such as LDOCE.
cmp-lg/9806016
Using WordNet for Building WordNets
cmp-lg cs.CL
This paper summarises a set of methodologies and techniques for the fast construction of multilingual WordNets. The English WordNet is used in this approach as a backbone for Catalan and Spanish WordNets and as a lexical knowledge resource for several subtasks.
cmp-lg/9806017
Anchoring a Lexicalized Tree-Adjoining Grammar for Discourse
cmp-lg cs.CL
We here explore a ``fully'' lexicalized Tree-Adjoining Grammar for discourse that takes the basic elements of a (monologic) discourse to be not simply clauses, but larger structures that are anchored on variously realized discourse cues. This link with intra-sentential grammar suggests an account for different patterns of discourse cues, while the different structures and operations suggest three separate sources for elements of discourse meaning: (1) a compositional semantics tied to the basic trees and operations; (2) a presuppositional semantics carried by cue phrases that freely adjoin to trees; and (3) general inference, that draws additional, defeasible conclusions that flesh out what is conveyed compositionally.
cmp-lg/9806018
Never Look Back: An Alternative to Centering
cmp-lg cs.CL
I propose a model for determining the hearer's attentional state which depends solely on a list of salient discourse entities (S-list). The ordering among the elements of the S-list covers also the function of the backward-looking center in the centering model. The ranking criteria for the S-list are based on the distinction between hearer-old and hearer-new discourse entities and incorporate preferences for inter- and intra-sentential anaphora. The model is the basis for an algorithm which operates incrementally, word by word.
cmp-lg/9806019
An Empirical Investigation of Proposals in Collaborative Dialogues
cmp-lg cs.CL
We describe a corpus-based investigation of proposals in dialogue. First, we describe our DRI compliant coding scheme and report our inter-coder reliability results. Next, we test several hypotheses about what constitutes a well-formed proposal.
cmp-lg/9806020
Textual Economy through Close Coupling of Syntax and Semantics
cmp-lg cs.CL
We focus on the production of efficient descriptions of objects, actions and events. We define a type of efficiency, textual economy, that exploits the hearer's recognition of inferential links to material elsewhere within a sentence. Textual economy leads to efficient descriptions because the material that supports such inferences has been included to satisfy independent communicative goals, and is therefore overloaded in Pollack's sense. We argue that achieving textual economy imposes strong requirements on the representation and reasoning used in generating sentences. The representation must support the generator's simultaneous consideration of syntax and semantics. Reasoning must enable the generator to assess quickly and reliably at any stage how the hearer will interpret the current sentence, with its (incomplete) syntax and semantics. We show that these representational and reasoning requirements are met in the SPUD system for sentence planning and realization.
cmp-lg/9807001
Evaluating a Focus-Based Approach to Anaphora Resolution
cmp-lg cs.CL
We present an approach to anaphora resolution based on a focusing algorithm, and implemented within an existing MUC (Message Understanding Conference) Information Extraction system, allowing quantitative evaluation against a substantial corpus of annotated real-world texts. Extensions to the basic focusing mechanism can be easily tested, resulting in refinements to the mechanism and resolution rules. Results are compared with the results of a simpler heuristic-based approach.
cmp-lg/9807002
The Role of Verbs in Document Analysis
cmp-lg cs.CL
We present results of two methods for assessing the event profile of news articles as a function of verb type. The unique contribution of this research is the focus on the role of verbs, rather than nouns. Two algorithms are presented and evaluated, one of which is shown to accurately discriminate documents by type and semantic properties, i.e. the event profile. The initial method, using WordNet (Miller et al. 1990), produced multiple cross-classification of articles, primarily due to the bushy nature of the verb tree coupled with the sense disambiguation problem. Our second approach using English Verb Classes and Alternations (EVCA) Levin (1993) showed that monosemous categorization of the frequent verbs in WSJ made it possible to usefully discriminate documents. For example, our results show that articles in which communication verbs predominate tend to be opinion pieces, whereas articles with a high percentage of agreement verbs tend to be about mergers or legal cases. An evaluation is performed on the results using Kendall's Tau. We present convincing evidence for using verb semantic classes as a discriminant in document classification.
cmp-lg/9807003
Centering in Dynamic Semantics
cmp-lg cs.CL
Centering theory posits a discourse center, a distinguished discourse entity that is the topic of a discourse. A simplified version of this theory is developed in a Dynamic Semantics framework. In the resulting system, the mechanism of center shift allows a simple, elegant analysis of a variety of phenomena involving sloppy identity in ellipsis and ``paycheck pronouns''.
cmp-lg/9807004
Word Clustering and Disambiguation Based on Co-occurrence Data
cmp-lg cs.CL
We address the problem of clustering words (or constructing a thesaurus) based on co-occurrence data, and using the acquired word classes to improve the accuracy of syntactic disambiguation. We view this problem as that of estimating a joint probability distribution specifying the joint probabilities of word pairs, such as noun verb pairs. We propose an efficient algorithm based on the Minimum Description Length (MDL) principle for estimating such a probability distribution. Our method is a natural extension of those proposed in (Brown et al 92) and (Li & Abe 96), and overcomes their drawbacks while retaining their advantages. We then combined this clustering method with the disambiguation method of (Li & Abe 95) to derive a disambiguation method that makes use of both automatically constructed thesauruses and a hand-made thesaurus. The overall disambiguation accuracy achieved by our method is 85.2%, which compares favorably against the accuracy (82.4%) obtained by the state-of-the-art disambiguation method of (Brill & Resnik 94).
cmp-lg/9807005
Graph Interpolation Grammars as Context-Free Automata
cmp-lg cs.CL
A derivation step in a Graph Interpolation Grammar has the effect of scanning an input token. This feature, which aims at emulating the incrementality of the natural parser, restricts the formal power of GIGs. This contrasts with the fact that the derivation mechanism involves a context-sensitive device similar to tree adjunction in TAGs. The combined effect of input-driven derivation and restricted context-sensitiveness would be conceivably unfortunate if it turned out that Graph Interpolation Languages did not subsume Context Free Languages while being partially context-sensitive. This report sets about examining relations between CFGs and GIGs, and shows that GILs are a proper superclass of CFLs. It also brings out a strong equivalence between CFGs and GIGs for the class of CFLs. Thus, it lays the basis for meaningfully investigating the amount of context-sensitiveness supported by GIGs, but leaves this investigation for further research.