SEARCH · Searcharxiv

Results for “cmp-lg”

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18Linked to original sources

arXiv1995

Bayesian Grammar Induction for Language Modeling

We describe a corpus-based induction algorithm for probabilistic context-free grammars. The algorithm employs a greedy heuristic search within a Bayesian framework, and a post-pass using the Inside-Outside algorithm. We compare the performance of our algorithm to n-gram models and the Inside-Outside algorithm in three language modeling tasks. In two of the tasks, the training data is generated by a probabilistic context-free grammar and in both tasks our algorithm outperforms the other techniques. The third task involves naturally-occurring data, and in this task our algorithm does not perform as well as n-gram models but vastly outperforms the Inside-Outside algorithm.

cmp-lg↗

arXiv1995

Response Generation in Collaborative Negotiation

In collaborative planning activities, since the agents are autonomous and heterogeneous, it is inevitable that conflicts arise in their beliefs during the planning process. In cases where such conflicts are relevant to the task at hand, the agents should engage in collaborative negotiation as an attempt to square away the discrepancies in their beliefs. This paper presents a computational strategy for detecting conflicts regarding proposed beliefs and for engaging in collaborative negotiation to resolve the conflicts that warrant resolution. Our model is capable of selecting the most effective aspect to address in its pursuit of conflict resolution in cases where multiple conflicts arise, and of selecting appropriate evidence to justify the need for such modification. Furthermore, by capturing the negotiation process in a recursive Propose-Evaluate-Modify cycle of actions, our model can successfully handle embedded negotiation subdialogues.

cmp-lg↗

arXiv1995

New Techniques for Context Modeling

We introduce three new techniques for statistical language models: extension modeling, nonmonotonic contexts, and the divergence heuristic. Together these techniques result in language models that have few states, even fewer parameters, and low message entropies. For example, our techniques achieve a message entropy of 1.97 bits/char on the Brown corpus using only 89,325 parameters. In contrast, the character 4-gram model requires more than 250 times as many parameters in order to achieve a message entropy of only 2.47 bits/char. The fact that our model performs significantly better while using vastly fewer parameters indicates that it is a better probability model of natural language text.

cmp-lg↗

arXiv1995

DATR Theories and DATR Models

Evans and Gazdar introduced DATR as a simple, non-monotonic language for representing natural language lexicons. Although a number of implementations of DATR exist, the full language has until now lacked an explicit, declarative semantics. This paper rectifies the situation by providing a mathematical semantics for DATR. We present a view of DATR as a language for defining certain kinds of partial functions by cases. The formal model provides a transparent treatment of DATR's notion of global context. It is shown that DATR's default mechanism can be accounted for by interpreting value descriptors as families of values indexed by paths.

cmp-lg↗

arXiv1995

Learning Syntactic Rules and Tags with Genetic Algorithms for Information Retrieval and Filtering: An Empirical Basis for Grammatical Rules

The grammars of natural languages may be learned by using genetic algorithms that reproduce and mutate grammatical rules and part-of-speech tags, improving the quality of later generations of grammatical components. Syntactic rules are randomly generated and then evolve; those rules resulting in improved parsing and occasionally improved retrieval and filtering performance are allowed to further propagate. The LUST system learns the characteristics of the language or sublanguage used in document abstracts by learning from the document rankings obtained from the parsed abstracts. Unlike the application of traditional linguistic rules to retrieval and filtering applications, LUST develops grammatical structures and tags without the prior imposition of some common grammatical assumptions (e.g., part-of-speech assumptions), producing grammars that are empirically based and are optimized for this particular application.

cmp-lg↗

arXiv1995

Treating Coordination with Datalog Grammars

In previous work we studied a new type of DCGs, Datalog grammars, which are inspired on database theory. Their efficiency was shown to be better than that of their DCG counterparts under (terminating) OLDT-resolution. In this article we motivate a variant of Datalog grammars which allows us a meta-grammatical treatment of coordination. This treatment improves in some respects over previous work on coordination in logic grammars, although more research is needed for testing it in other respects.

cmp-lg↗

arXiv1995

Evaluation of Semantic Clusters

Semantic clusters of a domain form an important feature that can be useful for performing syntactic and semantic disambiguation. Several attempts have been made to extract the semantic clusters of a domain by probabilistic or taxonomic techniques. However, not much progress has been made in evaluating the obtained semantic clusters. This paper focuses on an evaluation mechanism that can be used to evaluate semantic clusters produced by a system against those provided by human experts.

cmp-lg↗

arXiv1995

A Symbolic and Surgical Acquisition of Terms through Variation

Terminological acquisition is an important issue in learning for NLP due to the constant terminological renewal through technological changes. Terms play a key role in several NLP-activities such as machine translation, automatic indexing or text understanding. In opposition to classical once-and-for-all approaches, we propose an incremental process for terminological enrichment which operates on existing reference lists and large corpora. Candidate terms are acquired by extracting variants of reference terms through {\em FASTR}, a unification-based partial parser. As acquisition is performed within specific morpho-syntactic contexts (coordinations, insertions or permutations of compounds), rich conceptual links are learned together with candidate terms. A clustering of terms related through coordination yields classes of conceptually close terms while graphs resulting from insertions denote generic/specific relations. A graceful degradation of the volume of acquisition on partial initial lists confirms the robustness of the method to incomplete data.

cmp-lg↗

arXiv1995

Combining Multiple Knowledge Sources for Discourse Segmentation

We predict discourse segment boundaries from linguistic features of utterances, using a corpus of spoken narratives as data. We present two methods for developing segmentation algorithms from training data: hand tuning and machine learning. When multiple types of features are used, results approach human performance on an independent test set (both methods), and using cross-validation (machine learning).

cmp-lg↗

arXiv1995

The Compactness of Construction Grammars

We present an argument for {\em construction grammars} based on the minimum description length (MDL) principle (a formal version of the Ockham Razor). The argument consists in using linguistic and computational evidence in setting up a formal model, and then applying the MDL principle to prove its superiority with respect to alternative models. We show that construction-based representations are at least an order of magnitude more compact that the corresponding lexicalized representations of the same linguistic data. The result is significant for our understanding of the relationship between syntax and semantics, and consequently for choosing NLP architectures. For instance, whether the processing should proceed in a pipeline from syntax to semantics to pragmatics, and whether all linguistic information should be combined in a set of constraints. From a broader perspective, this paper does not only argue for a certain model of processing, but also provides a methodology for determining advantages of different approaches to NLP.

cmp-lg↗

arXiv1995

Context and ontology in understanding of dialogs

We present a model of NLP in which ontology and context are directly included in a grammar. The model is based on the concept of {\em construction}, consisting of a set of features of form, a set of semantic and pragmatic conditions describing its application context, and a description of its meaning. In this model ontology is embedded into the grammar; e.g. the hierarchy of {\it np} constructions is based on the corresponding ontology. Ontology is also used in defining contextual parameters; e.g. $\left[ current\_question \ time(\_) \right] $. A parser based on this model allowed us to build a set of dialog understanding systems that include an on-line calendar, a banking machine, and an insurance quote system. The proposed approach is an alternative to the standard "pipeline" design of morphology-syntax-semantics-pragmatics; the account of meaning conforms to our intuitions about compositionality, but there is no homomorphism from syntax to semantics.

cmp-lg↗

arXiv1995

User-Defined Nonmonotonicity in Unification-Based Formalisms

A common feature of recent unification-based grammar formalisms is that they give the user the ability to define his own structures. However, this possibility is mostly limited and does not include nonmonotonic operations. In this paper we show how nonmonotonic operations can also be user-defined by applying default logic (Reiter 1980) and generalizing previous results on nonmonotonic sorts (Young & Rounds 1993).

cmp-lg↗

arXiv1995

Semantic Ambiguity and Perceived Ambiguity

I explore some of the issues that arise when trying to establish a connection between the underspecification hypothesis pursued in the NLP literature and work on ambiguity in semantics and in the psychological literature. A theory of underspecification is developed `from the first principles', i.e., starting from a definition of what it means for a sentence to be semantically ambiguous and from what we know about the way humans deal with ambiguity. An underspecified language is specified as the translation language of a grammar covering sentences that display three classes of semantic ambiguity: lexical ambiguity, scopal ambiguity, and referential ambiguity. The expressions of this language denote sets of senses. A formalization of defeasible reasoning with underspecified representations is presented, based on Default Logic. Some issues to be confronted by such a formalization are discussed.

cmp-lg↗

arXiv1995

Development of a Spanish Version of the Xerox Tagger

This paper describes work performed withing the CRATER ({\em C}orpus {\em R}esources {\em A}nd {\em T}erminology {\em E}xt{\em R}action, MLAP-93/20) project, funded by the Commission of the European Communities. In particular, it addresses the issue of adapting the Xerox Tagger to Spanish in order to tag the Spanish version of the ITU (International Telecommunications Union) corpus. The model implemented by this tagger is briefly presented along with some modifications performed on it in order to use some parameters not probabilistically estimated. Initial decisions, like the tagset, the lexicon and the training corpus are also discussed. Finally, results are presented and the benefits of the {\em mixed model} justified.

cmp-lg↗

arXiv1995

Integrating Gricean and Attentional Constraints

This paper concerns how to generate and understand discourse anaphoric noun phrases. I present the results of an analysis of all discourse anaphoric noun phrases (N=1,233) in a corpus of ten narrative monologues, where the choice between a definite pronoun or phrasal NP conforms largely to Gricean constraints on informativeness. I discuss Dale and Reiter's [To appear] recent model and show how it can be augmented for understanding as well as generating the range of data presented here. I argue that integrating centering [Grosz et al., 1983] [Kameyama, 1985] with this model can be applied uniformly to discourse anaphoric pronouns and phrasal NPs. I conclude with a hypothesis for addressing the interaction between local and global discourse processing.

cmp-lg↗

arXiv1995

Ubiquitous Talker: Spoken Language Interaction with Real World Objects

Augmented reality is a research area that tries to embody an electronic information space within the real world, through computational devices. A crucial issue within this area, is the recognition of real world objects or situations. In natural language processing, it is much easier to determine interpretations of utterances, even if they are ill-formed, when the context or situation is fixed. We therefore introduce robust, natural language processing into a system of augmented reality with situation awareness. Based on this idea, we have developed a portable system, called the Ubiquitous Talker. This consists of an LCD display that reflects the scene at which a user is looking as if it is a transparent glass, a CCD camera for recognizing real world objects with color-bar ID codes, a microphone for recognizing a human voice and a speaker which outputs a synthesized voice. The Ubiquitous Talker provides its user with some information related to a recognized object, by using the display and voice. It also accepts requests or questions as voice inputs. The user feels as if he/she is talking with the object itself through the system.

cmp-lg↗

arXiv1995

Generating efficient belief models for task-oriented dialogues

We have shown that belief modelling for dialogue can be simplified if the assumption is made that the participants are cooperating, i.e., they are not committed to any goals requiring deception. In such domains, there is no need to maintain individual representations of deeply nested beliefs; instead, three specific types of belief can be used to summarize all the states of nested belief that can exist about a domain entity. Here, we set out to design a ``compiler'' for belief models. This system will accept as input a description of agents' interactions with a task domain expressed in a fully-expressive belief logic with non-monotonic and temporal extensions. It generates an operational belief model for use in that domain, sufficient for the requirements of cooperative dialogue, including the negotiation of complex domain plans. The compiled model incorporates the belief simplification mentioned above, and also uses a simplified temporal logic of belief based on the restricted circumstances under which beliefs can change. We shall review the motivation for creating such a system, and introduce a general procedure for taking a logical specification for a domain and procesing it into an operational model. We shall then discuss the specific changes that are made during this procedure for limiting the level of abstraction at which the concepts of belief nesting, default reasoning and time are expressed. Finally we shall go through a worked example relating to the Map Task, a simple cooperative problem-solving exercise.

cmp-lg↗

arXiv1995

Text Chunking using Transformation-Based Learning

Eric Brill introduced transformation-based learning and showed that it can do part-of-speech tagging with fairly high accuracy. The same method can be applied at a higher level of textual interpretation for locating chunks in the tagged text, including non-recursive ``baseNP'' chunks. For this purpose, it is convenient to view chunking as a tagging problem by encoding the chunk structure in new tags attached to each word. In automatic tests using Treebank-derived data, this technique achieved recall and precision rates of roughly 92% for baseNP chunks and 88% for somewhat more complex chunks that partition the sentence. Some interesting adaptations to the transformation-based learning approach are also suggested by this application.

cmp-lg↗