This papers presents a semantic role labeling system for Romanian texts. The semantic labeling system was developed using PASRL, a platform for supervised learning techniques. The developed platform tests several classifiers on different sub-problems of the SRL task (Predicate Identification, Predicate Sense Identification, Sense Identification, Argument Identification), chooses the ones with the greatest performance and returns a Semantic Role Labeling System (a sequence of trained models to run on new data).

Shallow semantic parsing of natural language processing is an important component in all kind of NLP applications and Semantic Role Labeling in particular, is an active research topic. This paper describes a rule-based Semantic Role Labeling system aimed at extracting semantic information from texts. The input text is processed by exploiting part of speech information and syntactic dependencies in order to identify semantic roles. The system's architecture is presented and the results and further developments are discussed.

This paper presents a semantic role labeling system acting as the backbone of an application that establishes the roles that entities have in different contexts, and what are the temporal, modal or local constraint that determine or restrict an event to take place. A semantic role represents the relationship between a predicate and an argument. Semantic parsing, by identifying and classifying the semantic entities in context and the relations between them, has great potential on its downstream applications, such as text summarization or machine translation.

This paper proposes a method to automatically improve a web page's metadata using the semantic content of the page. Thus, using a semantic role labeling system, the web page content is parsed and the entities that frequently play core semantic roles are considered for addition to the web page's list of metadata. Semantic role analysis answers questions such as: "What role has an entity in a specific context?" or "When, why, where or how an event takes place?".

This paper presents a semantic role labeling system acting as the backbone of an application that monitors the contexts and relations in which a specified entity appears in written texts. The main goal is to identify the role this entity plays, as well as the roles played by words frequently co-occurring with the input entity.

In order to develop a semantic labeling system, the most common methods use supervised learning from an annotated corpus. What if we have short deadlines and limited human and financial possibilities that prevent us from building such a training corpus for our language? If such a corpus already exists for any other language, this paper proposes a method to automatically import the existing corpus for the language we need. The transfer method is based on translating the existing corpus (or using annotated versions of existing parallel texts), aligning it at word level, and applying a set of mapping functions to import the annotation from one language to another. An import validation interface is also offered for the manual validation of the resulted resource. As an example, the case of semantic role import from the English FrameNet to Romanian is discussed.

The system presented in this paper creates a map of semantic information about a specific named entity (Person, Organization, etc.). Thus, after the user specifies a named entity, the system searches on the web and returns the first 200 web pages containing the specified entity, applies semantic roles on the returned paragraphs, and extracts a map of related actions involving the searched entity. This map of actions can then be chronologically ordered, thus illustrating the actions a certain entity has performed in a specific time frame (or at least the way it is reflected by the online world).

This paper describes preliminary analysis on the influence of the semantic roles in summary generation. The proposed method involves three steps: first, the named entities in the original text are identified using a named entity recognizer; secondly, the sentences are parsed and semantic roles are extracted; thirdly, selection of the sentences containing specific semantic roles for the most relevant entities in text. Although the method is language independent, in order to check its viability, we tested the proposed approach for Romanian summaries.

This paper presents the following results: (a) On the basis of an extensive overview of the currently Information Structure (IS) theories, the first goal of our paper is to update the IS terminology for the three important IS dimensions: Givenness, Background-Focus (also referred as Topic-Focus or Backgroud-Kontrast), and Topic-Comment (also Theme-Rheme). (b) We propose an intonational discourse-level hierarchy among the Contrastive Focus (First Occurrence Focus), Second Occurrence Focus, Informational (Discourse-New) Focus, and Deaccented (Discourse-Given) Focus, while the phonetic properties of the considered intonational inequalities remain to be statistically established and weighted through speech analysis for Romanian. (c) This discourse-level prosodic hierarchy is combined, in a separate and independent way, with the clause- and phrase-level intonational hierarchies driven by Sentence Accent Assignment Rules, Nuclear Stress Rule, and the more recently Sentence Break Assignment Rules. (d) Based on the intonational focus hierarchies at points (b) and (c) above, a new architecture for the Discourse-Prosody interface is outlined, aiming to replace the previous approaches of Topic-Focus and Theme-Rheme algorithms (which can provide only incomplete information) for prosody prediction of Romanian. (e) The notions of explicit and implicit contrastive focus are defined, and the meaningful relevance of the contrastive intonation for the Romanian finite-clauses is pointed out by significant percentages of the phenomena on George Orwell's "1984" corpus. (f) Classes of examples illustrate and evaluate, for Romanian, the intonational-prosodic patterns of the contrastive and non-contrastive focus markers, categories, and domains.

Classical methods of Information Structure computing, such as Topic-Focus Articulation, Theme-Rheme algorithms, or Segmented Discourse Representation Theory, have been considered for the intonational focus assignment on the Romanian discourse-prosody interface. The present paper proposes a more general, contrastivity-oriented architecture: language-dependent, specifically weighted, pre-established hierarchies of contrastive and non-contrastive intonational foci and breaks are coupled on discourse and clause-level textual structures and entities, aiming to merge and compete the classical approaches for an improved prosody prediction of Romanian.

This paper presents a Named Entity Recognition system for Romanian, created using linguistic grammar-based techniques and a set of resources. Our system's architecture is based on two modules, the named entity identification and the named entity classification module. After the named entity candidates are marked for each input text, each candidate is classified into one of the considered categories, such as Person, Organization, Place, Country, etc. The system's Upper Bound and its performance in real context are evaluated for each of the two modules (identification and classification) and for each named entity type. The evaluation show promising results, our system being comparable with the existing systems for Romanian, and even better for Person recognition.

This paper describes UAIC1's Question Answering for Machine Reading Evaluation systems participating in the QA4MRE 2011 evaluation task. The system is designed to extract knowledge from large volumes of text and to use this knowledge to answer questions in Romanian and English monolingual tasks. Our systems were built on the architecture of a Question Answering system, customized for this new task. Thus, the new system used from our previous question answering systems the question processing and information retrieval components, adapted for new requests. Additionally, a new component was added in order to detect the most probable answer of a question, from a list of possible answers.

This paper describes the preliminary results of a system for extracting sentiment and named entities. It also combines rule-based classification, statistics and machine learning in a new method. The accuracy and speed of extraction and classification are crucial. The service oriented architecture permits the end-user to work with a flexible interface in order to produce applications that range from aggregating consumer feedback on commercial products to measuring public opinion on political issues, from blog and forums. The experiment has two versions available for testing, one with concrete extraction results and sentiment calculus and the one with internal metrics validation results.

Semantic parsing, by identifying and classifying semantic entities in context, as well as the relations between them, has a great potential for applications such as text summarization, question-answering or machine translation. Thus, by developing a system that automatically annotates semantic roles for the Romanian language, this paper represents an important intermediate step towards autom atic natural language understanding. For the creation of the semantic role parser for Romanian, it was first necessary to develop a training corpus, annotated with semantic roles. This annotated resource of semantic roles for the Romanian language was created using as starting point a resource developed for English (FrameNet), and applying an automatic transfer method. Subsequently, using a platform for developing supervised semantic role labeling systems (PASRL), several learning algorithms were trained on the developed corpus and the best obtained model was saved. We discuss the results of applying this technique for Romanian, with the conviction that other languages could also benefit form using the same approach.

The next step in human-computer interaction is the use of human languages instead of some pre-defined commands. This paper intends to propose the use of natural language technologies in the management of industrial process flow. In order to make computers understand humans, language models need to be created from knowledge bases. We used such a knowledge base for the creation of a tool designed to analyse natural language semantics and determine who must perform what task where when etc. and we have the conviction that it can be used to make machines understand human when commanding the pipelining of processes.

This paper describes UAIC's Question Answering systems participating in the ResPubliQA 2010 competition, designed to answer questions on a juridical corpora in Romanian, English and French monolingual tasks. Our systems adhere to the classical architecture of a Question Answering system, with an emphasis on simplicity and real time answers: only shallow parsing was used for question processing, the indexes for the retrieval module were built at coarse-grained paragraph level, and the answer extraction component used simple pattern-based rules and lexical similarity metrics for candidate answer ranking.

2009 marked UAIC1's fourth consecutive participation at the QA@CLEF competition, with continually improving results. This paper describes UAIC's QA systems participating in the Ro-Ro and En-En tasks. Both systems adhered to the classical QA architecture, with an emphasis on simplicity and real time answers: only shallow parsing was used for question processing, the indexes used by the retrieval module were at coarse-grained paragraph and document levels, and the answer extraction component used simple pattern-based rules and lexical similarity metrics for candidate answer ranking. The results obtained for this year's participation were greatly improved from those of our team's previous participations, with an accuracy of 54% on the EN-EN task and 47% on the RO-RO task.

This paper presents a cross-linguistic analysis of the largest dictionaries currently existing for Romanian, French, and German, and a new, robust and portable method for Dictionary Entry Parsing (DEP), based on Segmentation-Cohesion-Dependency (SCD) configurations. The SCD configurations are applied successively on each dictionary entry to identify its lexicographic segments (the first SCD configuration), to extract its sense tree (the second configuration), and to parse its atomic sense definitions (the third one). Using previous results on DLR (The Romanian Thesaurus - new format), the present paper adapts and applies the SCD-based technology to other four large and complex thesauri: DAR (The Romanian Thesaurus - old format), TLF (Le Trésor de la Langue Française), DWB (Deutsches Woerterbuch - GRIMM), and GWB (Goethe-Woerterbuch). This experiment is illustrated on significantly large parsed entries of these thesauri, and proved the following features: (1) the SCD-based method is a completely formal grammar-free approach for dictionary parsing, with efficient (weeks-time adaptable) modeling through sense hierarchies and parsing portability for a new dictionary. (2) SCD-configurations separate and run sequentially and independently the processes of lexicographic segment recognition, sense tree extraction, and atomic definition parsing. (3) The whole DEP process with SCD-configurations is optimal. (4) SCD-configurations, through sense marker classes and their dependency hypergraphs, offer an unique instrument of lexicon construction comparison, sense concept design and DEP standardization.

This paper presents a lexical-semantics cross-linguistic analysis of the largest thesauri currently existing for Romanian, French, and German, and a new, robust and portable method for their entry parsing, based on the technique of Segmentation-Cohesion-Dependency (SCD) configurations. The general idea behind parsing a thesaurus or dictionary is to transform a raw text dictionary entry into an indexable linguistic database of lexical-semantics (sense) trees. The SCD parsing configurations are applied successively on a thesaurus entry in order to identify its lexicographic segments (the first SCD configuration), to extract the tree of its senses and subsenses (the second one), and to parse its atomic and non-atomic sense definitions (the third one). Using previous results on DLR (The Romanian Thesaurus - new format), the present paper adapts and applies the SCD-based technology to other four large and complex thesauri: DAR (The Romanian Thesaurus - old format), TLF (Le Trésor de la Langue Française), DWB (Deutsches Wörterbuch - GRIMM), and GWB (Göthe-Wörterbuch). This experiment, which was illustrated on significantly large parsed entries of the mentioned thesauri, proved the main achievements of the parsing technique based on SCD-configurations: efficiency, robustness, portability. These qualities, when compared with the classical methods for dictionary entry parsing, derive mainly from at least two SCD-specific facts: the sense tree extraction is performed on the sense marker sequences exclusively, while the processes of sense tree extraction and (atomic) sense definition parsing are shown to be completely separable.

The present paper, maintaining the interest for applying Prague School's Topic-Focus Articulation (TFA) algorithm to Romanian, takes the advantage of an experiment of investigating the intonational focus assignment to the Romanian sentence. Using two lines of research in a previous study, it has been showed that TFA behaves better than an inter-clausal selection procedure for assigning pitch accents to the Background-Kontrast (Topic-Focus) entities, while the Inference Boundary algorithm for computing the Theme-Rheme is more reliable and easier extendable towards boundary and contour tone assignment rules, leading to our novel proposal of Sentence Boundary Assignment Rules (SBAR). The main contributions of this paper are: (a) The TFA algorithm applied for Romanian is extended to inter-clause level, and embedded into a discursive approach for computing the Background-Kontrast entities. (b) Inference Boundary algorithm, applied to Romanian for clause-level Theme-Rheme span computing, is extended with a set of SBAR rules, which are relying on different Communicative Dynamism (CD) degrees of the clause constituents. (c) On each intonational unit, the extended TFA algorithm is further refined, in order to eliminate ambiguities, with a filter derived from Gussenhoven's SAAR (Sentence Accent Assignment Rule). Remarkably, TFA and CD degrees proved again to be resourceful, especially as technical procedures for a new and systematic development of an Information Structure Discourse Theory (ISDT).

"...it is not the argument structure that triggers the intonational phrasing, but the [discourse, subclausal n.b.] relation of backgrounding." (K. von Heusinger, 2007) Is it? If yes, how? The present paper, maintaining our attempts for applying Prague School's Topic-Focus Articulation (TFA) algorithm on the syntax-prosody interface of Romanian, proposes two comparative lines of investigation for the intonational focus assignment: (A) The TFA algorithm is improved at clause level with hints from Gussenhoven's SAAR (Sentence Accent Assignment Rule), then extended to inter-clause level, i.e. complex sentences. The new shape of the TFA algorithm is applied to compute the Topic-Focus values in the discursive context, while the information-structural (IS) spans of Theme(s)-Rheme(s) are detached, at clause level, as lowest-highest degrees of Communicative Dynamism (CD) vs. Systemic Order (SO). (B) The second approach we experiment for assigning intonational focus and phrasing is based on the combined and intensive use of discourse theories for computing the IS categories and structures: the Background-Kontrast entities (associated with the Prague School's Topic-Focus) are obtained with Asher's (1993) Segmented Discourse Representation Theory (SDRT) analysis, while Theme-Rheme structures within the finite clause are computed with Leong's (2004) Inference-Boundary (IB) algorithm (of Hallidayan inspiration), applied for the first time to Romanian. Furthermore, this second direction is inspired from and joins the IS-discourse theory proposed by Heusinger (2007), which relies on SDRT inter-clausal evolution of discourse variables for computing the Background-Kontrast. While maintaining the classical SDRT (including rhetorical) discourse relations at the inter-clause level, Heusinger introduces a set of IS-semantics relations, and hands down at sub-clause level the rhetorical and focus particle relations with significant role in intonational-prosodic phrasing. Examples of these two types of research are compared to a gold, intonationally annotated set of Romanian sentences, the proposed theoretical and procedural techniques aiming to balance the pessimistic-realistic view on prosody prediction that it is the speaker-presupposition (and hearer-accommodation) which determines the IS focal scopes, rather than the bare text.

2008 marked UAIC's third consecutive participation at the QA@CLEF competition, with continually improving results. The most significant change to our system with regards to last year is the partial transition to a real-time QA system, as a consequence of the simplification or elimination of the main time-consuming tasks such as linguistic pre-processing. A brief description of our system and an analysis of the errors introduced by each module are given in this paper.

The paper presents a method to restore diacritics using web contexts. The system receives one or more sentences in one language and uses the Google engine to recover diacritics for the sentence words. The system accuracy is similar to the accuracy of existing systems, but the main advantage comes from fact that it uses resource and tools available for free or that are easy to obtain for other languages, leading us to believe that this approach could be valid for more languages.

This year marked UAIC's fourth consecutive participation at the QA@CLEF competition, with continually improving results. This year we participated successfully both in Ro-Ro task and in En-En task pf the ResPubliQA track. A brief description of our system is given in this paper.

This paper aims to introduce a new parsing strategy for large dictionary (thesauri) parsing, called Dictionary Sense Segmentation and Dependency (DSSD), devoted to obtain the sense tree, i.e. the hierarchy of the defined meanings, for a dictionary entry. The real novelty of the proposed approach is that, contrary to dictionary 'standard' parsing, DSSD looks for and succeeds to separate the two essential processes within a dictionary entry parsing: sense tree construction and sense definition parsing. The key tools to accomplish the task of (autonomous) sense tree building consist in defining the dictionary sense marker classes, establishing a tree-like hierarchy of these classes, and using a proper searching procedure of sense markers within the DSSD parsing algorithm. A similar but more general approach, using the same techniques and data structures for (Romanian) free text parsing is SCD (Segmentation-Cohesion-Dependency), which DSSD is inspired from. A DSSD-based parser is implemented in Java, building currently 91% correct sense trees from DTLR (Dicţionarul Tezaur al Limbii Române - Romanian Language Thesaurus) entries, with significant resources to improve and enlarge the DTLR lexical semantics analysis.

The paper describes the creation a grammar developed to automatically extract definitions from web documents. Three evaluation scenarios were run, the results of these experiments being the main focus of the paper. The results are convincing, so that further development as well as further integration of the definition extraction system in various related applications are already under way.

This year marked UAIC's third consecutive participation at the QA@CLEF competition, with continually improving results. The most significant changes to our system with regards to last year is the partial transition to a real-time QA system, consequences being the simplification or elimination of principal time-consuming tasks such as linguistic pre-processing. A brief description of our system and an analysis of the errors introduced by each module are described in this paper.

This paper briefly describes UAIC's participation in this year's CLEF question answering competition, focusing on the main challenges and changes compared to our last year participation. An analysis of the errors introduced by each module is also discussed.

The double subject in Romanian sentences is a controversial linguistic phenomenon. While some researchers accept it as a language 'curiosity', others consider it apposition, in order to embody its behaviour in the already existing theories. We present the first study in the international literature on the phonetic analysis of double-subject sentences.

Semantic databases are a stable starting point in developing knowledge based systems. Since creating language resources demands many temporal, financial and human resources, a possible solution could be the import of a resource annotation from one language to another. This paper presents the creation of a semantic role database for Romanian, starting from the English FrameNet semantic resource. The intuition behind the import program is that most of the frames defined in the English FN are likely to be valid cross-lingual, since semantic frames express conceptual structures, language independent at the deep structure level. The surface level is realized according to each language's syntactic constraints. In the paper we present the advantages of choosing to import of the English FrameNet annotation, instead of annotating a new corpus. We also take into account the mismatches encountered in the validation process. The rules created to manage particular situations are used to improve the import program. We believe the information and argumentations in this paper could be of interest for those who wish develop FrameNet-like systems for other languages.

The aim of this paper is to investigate the syntactic / semantic substructures (called subgroups) of the Romanian verbal group (VG), or verbal complex, starting with the achievements in the literature, and melted into the device of direct and inverse functional projection within FX-bar theory. The paper examines several problems and their solutions for the syntactic-semantic theories of VG, as discussed in some fundamental papers, and we offer our explanation on the involved syntactic phenomena, the emphasis falling on the VG substructures (verbal subgroups, VSGs), VSG boundaries and composition within VG, direct and inverse FX-bar projections of VG, VG parsing, lexical semantics and intensional / extensional logic representations of the Romanian (verbal or nominal) predicate.

This paper proposes and discusses issues that are connected to local and global text structures, all of them being associated in a way or another to the concept of predication. Outcome of various stages of evolution (including our own) of X bar theory, the present work comprises the following topics: (a) A novel functional X bar (FX bar) scheme is proposed, aiming to reveal, model and relate the local, clause-level markers and text structures. (b) At global level, two FX bar schemes are proposed, one pursuing the inter-clause level relations (of logical, syntactic, and second-order theta-semantics nature), and the other being of discursive, rhetorical nature. (c) There are described local and global classes of markers, to be incorporated on the projection levels of FX bar schemes and within SCD (Segmentation-Cohesion-Dependency) linguistic strategy algorithms, together with the graph-based hierarchy among these classes. (d) The concept of functional generativity is discussed, with implications on parsing algorithm classification and FX bar projection mechanism. (e) Local FX bar projection functions have at their core the notion of lexical predication. Direct (towards clause) and inverse (towards lexicon) FX bar projections of the verbal group (verbal complex) are shown to balance between syntactic and semantic diatheses, taking steps for a better understanding of predicate and predication anatomies. (f) Finally, direct and inverse global FX bar projections mediate between larger text spans and inter-clause vs. discourse trees, the intricate relationship between the finite clause and (sub-clause and multi-clause) discourse segment being highlighted.

The following main ideas have been pointed out and put to work within our paper: (a) Information Structure (IS) theories (topic-focus, theme-rheme, background-contrast, informational and contrastive focus, focus projection rules, etc.) on text are shown to behave currently as a consistent linguistic tool that can stand behind a correct, language and contextual-depending, mapping of text into speech discourse, on the basis of local-global, textual-intonational structures and markers. (b) Functional and logical-semantics (including pragmatics) approaches for IS modelling of the local and global syntax and discourse, coming from the text side, and reliable intonational units within coherent theories on spoken discourse, coming from the prosody side, are needed. We outline here the SCD (Segmentation-Cohesion-Dependency) and FDG (Functional Dependency Grammar) parsing strategies and linguistic theories. (c) Concerning the IS semantics, this paper takes a first (experimental) step to design an adequate syntax-prosody interface for Romanian, by adapting and applying the Prague School's TFA (Topic-Focus Articulation) algorithm. The contributions we consider worth to be mentioned are: (d1) theoretical and computational local-global parsing frameworks of Romanian, emphasizing the functional and hierarchical organization of linguistic categories and markers: SCD and FDG. (d2) outline of the relationship between textual local-global syntactic-discourse structures and prosodic intonational units; (d3) implementing for the first time (to our knowledge) the TFA algorithm to Romanian.

We propose in this paper an implementation of the Prague School's TFA (Topic-Focus Articulation) algorithm to support the Romanian prosody design, relying on the experience with FDG (Functional Dependency Grammar) and SCD (Segmentation-Cohesion-Dependency) parsing strategies for the classical, i.e. predication-driven, but Information Structure (IS) non-dependent, syntax. As contributions worth to be mentioned are: (a) Outlining the functional and hierarchical organization of linguistic markers and structures within SCD and FDG local-global parsing, on both sides of the syntax-prosody interface of Romanian. (b) Pointing out the relationship between classical (IS-free) syntactic structures, IS (topic-focus, communicative dynamism) depending textual spans, and the corresponding prosodic intonational units. (c) Adapting andimplementing the TFA algorithm for the first time to Romanian prosodic structures, to be continued with TFA sentence-level refinements, its rhetorical-level extension, and embedding into local-global linking algorithms.

This paper presents part of our work in the LT4eL project regarding the grammar developed by the Romanian team in order to extract definitions from texts. Some qualitative results come in order to evaluate our grammar rules. Among the applications of this kind of grammar we will discuss the possible inclusion of the grammar rules into a question answering system in order to extract answers for definition type questions. Another possible usage of those rules envisages the extraction of supplementary knowledge from linguistic resources like Wikipedia. The benefits of such an extra-knowledge resource are evident in textual entailment systems, where some resources like WordNet, Acronyms database or Dirt cannot cover all the requirements of the system.

This paper describes the development of a Question Answering (QA) system and its evaluation results in the Romanian-English cross-lingual track organized as part of the CLEF 2006 campaign. The development stages of the cross-lingual Question Answering system are described incrementally throughout the paper, at the same time pinpointing the problems that occurred and the way they were addressed. The system adheres to the classical architecture for QA systems, debuting with question processing followed, after term translation, by information retrieval and answer extraction. Besides the common QA difficulties, the track posed some specific problems, such as the lack of a reliable translation engine from Romanian into English, and the need to evaluate each module individually for a better insight into the system's failures.

We propose a method based on spoken language analysis to deal with controversial syntactic issues; we apply the method to the problem of the double subject sentences in the Romanian language. The double subject construction is a controversial linguistic phenomenon in Romanian. While some researchers accept it as a language 'curiosity' (specific only to the Asian languages, but not to the European ones), others consider it apposition-type structure, in order to embody its behaviour in the already existing theories. This paper brings a fresh gleam of light over the debate, by presenting what we believe to be the first study on the phonetic analysis of double-subject sentences in order to account for its difference vs. the appositional constructions.

We present an initiative to develop an educational and scientific, Internet-based, freely accessible spoken language resource for the sounds and words of the Romanian language. The database contains several versions of all the basic sounds of the Romanian language, moreover sets of words. This language resource is suitable for education, moreover for statistical research on the Romanian spoken language, and for testing and benchmark purposes. The corpus is extensively documented, in order to allow both scientific research and informed education based on it. This corpus, currently located at the address, can be used for purposes such as analysis of sounds, analysis of specificities of the Romanian language pronunciation compared to other languages, and Romanian language learning aided by computer. The resource was created by the cooperation of teachers, researchers and Ph.D. students from three academic institutions.

We present a study of the prosody - seen in a broader sense - that supports the theory of the interrelationship function of speech. Both "pure emotions" and specific syntactical constructions, like the double-subject propositions, are meant to show a relationship of the speaker with the general context. The analysis goes beyond the basic prosody, as related to pitch values, pitch trajectory, sound duration and pauses; the analysis also aims to determine the change in higher formants. The refinement of the analysis asks for finer tools. Methodological aspects are discussed, including limitations of the currently available tools.

The main contributions enclosed within this paper refer to: (i) A general view on the syntax-prosody interface, the non-isomorphism between the two facets of the language being conjectured as a proper subsumption mapping between syntax and prosody; (ii) Implementation of the Topic-Focus Articulation (TFA) algorithm for the Romanian sentence, with novelties on Intonational TFA and interrogative sentences.; (iii) Devising the verbal group (VG, viz. verbal complex) into verbal subgroup syntactic structures; (iv) TFA algorithm refinement on complex Romanian VGs; (v) Subjectivity / Objectivity discovery based on TFA and communicative dynamism disordering, with consequences for text analysis and prosody design in e learning systems.

Interests to realize semantic frame databases as a stable starting point in developing semantic knowledge based systems exists for languages such as German (the Salsa project), English (the PropBank project, the FrameNet project), Spain, Japanese, etc. I thus propose to create a semantic frame database for Romanian, similar to the FrameNet database. Since creating language resources demands many temporal, financial and human resources, a possible solution could be the import of standardized annotation of a resource developed for a specific language to other languages. This paper presents such a method for the import of the FrameNet annotation from English to Romanian.

The aim of this paper is two-fold: (a) to discuss the structure (and substructures) of the verbal complex, or verbal group (VG), investigated within the framework of functional FX bar (projection) theory, with the semantic instruments of intensional / extensional logic, and (b) to expose results on the VG parsing task, embodied into a more general FDG clause-level parser. We provide a unitary taxonomy of the verbal and nominal predicate, based on intensional logic. Both constructions rely on verbal subgroups (VSGs) that a VG is decomposed in, with tense auxiliary, copulative, modal, semi-auxiliary, restructuring head verbs. VSG substructures are (recursively) composing one another, using the predication feature (with valence-arity, argument type-sort, and polymorphism, if necessary) of their lexical and grammatical heads, to obtain complex FX bar projections representing the VG. The general description of the VG parsing program is given, its purpose being to extract from Romanian texts, VGs (or verbal predicates), together with their additional properties such as tense and voice (syntactic diathesis). The program is written in Java, and uses regular expressions for determining the basic and compositional properties of VGs.

The aim of this paper is to investigate the syntactic / semantic substructures of the Romanian verbal group (VG), or verbal complex, starting from the instruments and arguments in the literature, and melted into the device of (direct and inverse) FX bar projections and theory. The paper will examine several problems and their solutions for the syntactic theories of VG, as discussed in some fundamental papers, and we shall offer our explanation on the involved syntactic phenomena, the emphasis falling on the lexical semantics and intensional / extensional logic representations, our interests being mainly oriented towards VG parsing, VG substructures (verbal subgroups, VSGs), VSG composition and their direct and inverse FX bar projections.

We present the steps of the annotation process for some phrases of the Romanian Spoken Language. The recordings have been made from a set of young subjects and the results are presented.

This paper describes the development of a question answering system for the Romanian-English cross-lingual track organized within the Cross Lingual Evaluation Forum (CLEF) 2006 campaign. The development stages of our cross-lingual Question Answering (QA) system are described incrementally throughout the paper, at the same time pinpointing the problems that occurred and the way they were addressed. Our system adheres to the classical architecture for QA systems, debuting with question processing followed, after term translation, by information retrieval and answer extraction. Besides the common QA difficulties, the track posed some specific problems, such as the lack of a reliable translation engine from Romanian to English, and the need to evaluate each module individually for a better insight into the system's failures.

The aim of this paper is to deepen the discussion of the role of shimmer and jitter in the natural speech and to assess the importance of shimmer and jitter for the synthesized speech. Toward this goal, nonlinear analysis methods are applied.

The paper reports difficulties encountered during the alignment of synsets between English and Romanian. Reasons for these difficulties are inconsistencies found both in Princeton WordNet (we will refer to it from now on with PWN), on one part, and in our sources, on the other part, the difference in criteria based on which senses were recorded in PWN and in our sources, and the inherent differences in the lexicalization of concepts in the two languages.