Search

Resumen. En este trabajo examinamos el problema de clasificar la orientacion semantica de opiniones de productos comerciales, en idioma Espanol. Exploramos las caracteristicas de la coleccion de opiniones extraidas de Internet y estudiamos el uso de bigramas de afirmacion como caracteristicas de entrenamiento para un metodo supervisado. Analizamos la combinacion de estas caracteristicas y medimos su desempeno al utilizarse en el metodo de Maquinas de soporte vectorial. Los resultados se equiparan con el estado del arte para opiniones en espanol.

Being able to create a natural language compiler has been one of the most sought-after goals to reach since the very beginning of artificial intelligence. Since then; however, it has been an elusive and difficult task to achieve to the extent of being considered almost impossible to perform. In this article, we present a promising path by using a grammar formalism which attempts to model natural language; in principle, by using minimalist grammars as one of the last proposed instances of formalism of this type. The main idea consists in creating a parser based on this type of grammars which could recognize and analyze the text (or input program) written in natural language and use this parser as a front-end of a compiler. Then, for the rest of the compilation process, utilize the usual phases of a classic compiler of a programming language. Moreover, we present a prototype of a natural language compiler whose specific language is that of arithmetic expressions, in order to show with evidence that it is indeed possible to implement it, that is to say, to put the proposed compiler design into practice, showing in this manner that it is actually possible to create a natural language compiler following this promising path.

We present a linguistic analysis of Named Entities in Spanish texts. Our work is focused on the determination of the structure of complex proper names: names with coordinated constituents, names with prepositional phrases and names formed by several content words initialized by a capital letter. We present the analysis of circa 49,000 examples obtained from Mexican newspapers. We detailed their structure and give some notions about the context surrounding them. Since named entities belong to open class of words they are being created daily, so the challenge for a named entity recognizer is to precisely determine the boundaries of new entity names in any text and to analyze thoroughly their components for deep semantic analysis. Knowing their general classes of structure it should be possible to derive useful heuristics or a specific grammar for natural language processing applications.

We study the effect of negation cues on semantic orientation prediction. State-of-the-art approaches to semantic orientation derivation are based on automatic classification. We analyze the use of negation cues as features for both supervised and unsupervised methods. We apply such methods on a collection of washing-machine reviews in Spanish. We compare the results of the two approaches and discuss the performance of each negation cue. We found that simple features performed similarly to using more resources.

From semantic point of view, information is usually contained in small units, called facts that are usually smaller than sentences. Identification of these facts in a text is not a trivial task. We present a heuristic algorithm for extraction of facts from sentences using a simple representation based on a relational data model. We focus our study on texts that contain a lot of facts by their nature: structured textbooks. The algorithm is based on data obtained by a syntactic analyzer. The obtained facts can be useful for information retrieval tasks, automatic summarization, etc. Our experiments are conducted for Spanish language. We obtained better results than the similar methods.

This paper reports research on Spanish temporal expressions. The analyzed phrases include a common temporal expression for a period of years reinforced by an adverb of time. We found that some of those phrases are age-related expressions. We analyzed a sample obtained from the Internet to determine the local context where the age-related meaning is unmistakable. We present the results for 21 selected classes.

Cohesion between components of collocations is already acknowledged measurable by means of the Web, and cohesion measurements are used for some applications and extraction of new collocations. Taking a specific cohesion criterion SCI, we performed massive evaluations of collocate cohesion in Oxford Collocations Dictionary. For three groups of modificative collocations (adjective—noun, adverb—adjective, and adverb— verb) SCI distributions proved to be one-peaked and compact, with rather close mean values and standard deviations. Thus we suggest a reliable numeric criterion for extraction of collocations from the Web.

Stable coordinate pairs (SCP) like comentarios y sugerencias ‘comments and suggestions’ or sano y salvo ‘safe and sound’ are rather frequent in texts in Spanish, though there are only few thousands of them in language. We characterize SCPs statistically by a numerical Stable Connection Index and reveal its unimodal distribution. We also propose lexical, morphologic, syntactic, and semantic categories for SCP structural description — for both a whole SCP and its components. It is argued that database containing a set of categorized SCPs facilitates several tasks of automatic NLP.. The research is based on a set of ca. 2200 Spanish coordinate pairs.

The objective of this work is to automatically determine, in an unsupervised manner, Spanish prepositional phrases of the type preposition – nominal phrase – preposition (P(NP(P) that behave in a sentence as a lexical unit and their semantic and syntactic properties cannot be deduced from the corresponding properties of each simple form, e.g., por medio de (by means of), a fin de (in order to), con respecto a (with respect to). We show that idiomatic P(NP(P combinations have some statistical properties distinct from those of usual idiomatic collocations. We also explore a way to differentiate P(NP(P combinations that could perform either as a regular prepositional phrase or as idiomatic prepositional phrase.

Malapropisms are real-word errors that lead to syntactically correct but semantically implausible text. We report an experiment on detection and correction of Spanish malapropisms. Malapropos words semantically destroy collocations (syntactically connected word pairs) they are in. Thus we detect possible malapropisms as words that do not form semantically plausible collocations with neighboring words. As correction candidates, we select words similar to the suspected one but forming plausible collocations with neighboring words. To judge semantic plausibility of a collocation, we use Google statistics of occurrences of the word combination and of the two words taken apart. Since collocation components can be separated by other words in a sentence, Google statistics is gathered for the most probable distance between them. The statistics is recalculated to a specially defined Semantic Compatibility Index (SCI). Heuristic rules are proposed to signal malapropisms when SCI values are lower than a predetermined threshold and to retain a few highly SCI-ranked correction candidates. Our experiments gave promising results.