Manuel Vilares Ferro

Regional Least-Cost Error Repair

We describe an algorithm to deal with automatic error repair over unrestricted context-free languages. The method relies on a regional least-cost repair strategy with validation, gathering all relevant information in the context of the error location. The system guarantees the asymptotic equivalence with global repair strategies.

A Common Solution for Tokenization and Part-of-Speech Tagging

Current taggers assume that input texts are already tokenized, i.e. correctly segmented in tokens or high level information units that identify each individual component of the texts. This working hypothesis is unrealistic, due to the heterogeneous nature of the application texts and their sources. The greatest troubles arise when this segmentation is ambiguous. The choice of the correct segmentation alternative depends on the context, which is precisely what taggers study.In this work, we develop a tagger able not only to decide the tag to be assigned to every token, but also to decide whether some of them form or not the same term, according to different segmentation alternatives. For this task, we design an extension of the Viterbi algorithm able to evaluate streams of tokens of different lengths over the same structure. We also compare its time and space complexities with those of the classic and iterative versions of the algorithm.

Towards the Development of Heuristics for Automatic Query Expansion

In this paper we study the performance of linguistically motivated conflation techniques for Information Retrieval in Spanish. In particular, we have studied the application of productive derivational morphology for single word term conflation and the extraction of syntactic dependency pairs for multi-word term conflation. These techniques have been tested on several search engines implementing different indexing models. The aim of this study is to find the strong and weak points of each technique in order to develop heuristics for automatic query expansion.

Instrumentation of Synchronous Reactive Models for Performance Engineering

Synchronous Reactive Modelling provides an optimal framework for the modular decomposition of programs that engage in complex patterns of deterministic interaction, such as many real-time and communication entities. This paper presents an approach which includes performance modelling techniques in the Synchronous Reactive Modelling method supported by Esterel. It defines a methodology based on timing and probabilistic quantitative constructs which complete the functional models. A monitoring mechanism provides performance results during the simulation. This methodology is applied to a protocol modelling case study. Performance metrics are computed and compared with known reference results.

Approximate Pattern Matching in Shared-Forest

We present a proposal intended to demonstrate the applicability of tabulation techniques to pattern recognition problems, when dealing with structures sharing some common parts. This work in motivated by the study of information retrieval for textual databases, using pattern matching as a basis for querying data.

Approximate VLDC Pattern Matching in Shared-Forest

We present a matching-based proposal intended to deal with querying for structured text databases. Our approach extends approximate VLDC matching techniques by allowing a query to exploit sharing of common parts between patterns used to index the document.

Dynamic programming as frame for efficient parsing

The last few years have seen a renewal of interest in the consideration of dynamic programming in compiler technology. This is due to the compactness of the representations, which are turning this paradigm into a common way of dealing with highly redundant computations occurring, for instance, in natural language processing, logic programming or abstract interpretation and related to phenomena such as non determinism or domain ordering. However, it is not usual to find practical studies about what the real interest of these techniques is, and which approaches are better adapted in each case. We justify the consideration of dynamic programming, both in definite clause and context free grammar parsing, highlighting the parallelism between the conclusions reached in both cases. We focus on the computational properties, suggesting a simple decision guide for the reader interested in applying this technology.

Finite state morphology and formal verification

The full paper describes an environment for the generation of non-deterministic taggers, currently used for the development of a Spanish lexicon. In relation to previous approaches, our system includes the use of verification tools in order to assure the robustness of the generated taggers. A wide variety of user defined criteria can be applied for checking the exact properties of the system.

Practical NLP-Based Text Indexing

We consider a set of natural language processing techniques based on finite-state technology that can be used to analyze huge amounts of texts. These techniques include an advanced tokenizer, a part-of-speech tagger that can manage ambiguous streams of words, a system for conflating words by means of derivational mechanisms, and a shallow parser to extract syntactic-dependency pairs. We propose to use these techniques in order to improve the performance of standard indexing engines.