Rodolfo Delmonte

Building the Italian Syntactic-Semantic Treebank

The paper reports on the design and construction of a multi-layered corpus of Italian, annotated at the syntactic and lexico-semantic levels, whose development is supported by dedicated software augmented with an intelligent interface. The issue of evaluating this type of resource is also addressed.

SLIM prosodic automatic tools for self-learning instruction

Abstract We present the Prosodic Module of a courseware for computer-assisted foreign language learning called SLIM – an acronym for Multimedia Interactive Linguistic Software, developed at the University of Venice (see Delmonte et al., 1999a,b). The Prosodic Module has been created in order to deal with the problem of improving a students performance both in the perception and production of prosodic aspects of spoken language activities. It is composed of two different sets of Learning Activities, the first one dealing with phonetic and prosodic problems at word level and at segmental level – where segmental refers to syllable-sized segments; the second one dealing with prosodic aspects at phonological phrase and utterance suprasegmental level. The main goal of Prosodic Activities is to ensure consistent and pedagogically sound feedback to the student intending to improve his/her pronunciation in a foreign language. We argue that the use of Automatic Speech Recognition (ASR) as Teaching Aid should be under-utilized and should be targeted to narrowly focussed spoken exercises, disallowing open-ended dialogues, in order to ensure consistency of evaluation. In addition, we argue that ASR alone cannot be used to gauge Goodness of Pronunciation (GOP), being inherently inadequate for that goal. On the contrary, we support the conjoined use of ASR technology and prosodic tools to produce GOP useable for linguistically consistent and adequate feedback to the student.

Opinion Mining and Sentiment Analysis Need Text Understanding

We argue in this paper that in order to properly capture opinion and sentiment expressed in texts or dialogs any system needs a deep linguistic processing approach. As in other systems, we used ontology matching and concept search, based on standard lexical resources, but a natural language understanding system is still required to spot fundamental and pervasive linguistic phenomena. We implemented these additions to VENSES system and the results of the evaluation are compared to those reported in the state-of-the-art systems in sentiment analysis and opinion mining. We also provide a critical review of the current benchmark datasets as we realized that very often sentiment and opinion is not properly modeled.

Automatic argumentative analysis for interaction mining

Interaction mining is about discovering and extracting insightful information from digital conversations, namely those human–human information exchanges mediated by digital network technology. We present in this article a computational model of natural arguments and its implementation for the automatic argumentative analysis of digital conversations, which allows us to produce relevant information to build interaction business analytics applications overcoming the limitations of standard text mining and information retrieval technology. Applications include advanced visualisations and abstractive summaries.

Entailment and Anaphora Resolution in RTE3

We present VENSES, a linguistically-based approach for semantic inference which is built around a neat division of labour between two main components: a grammatically-driven subsystem which is responsible for the level of predicate-arguments well-formedness and works on the output of a deep parser that produces augmented head-dependency structures. A second subsystem fires allowed logical and lexical inferences on the basis of different types of structural transformations intended to produce a semantically valid meaning correspondence. In the current challenge, we produced a new version of the system, where we do away with grammatical relations and only use semantic roles to generate weighted scores. We also added a number of additional modules to cope with fine-grained inferential triggers which were not present in previous dataset. Different levels of argumenthood have been devised in order to cope with semantic uncertainty generated by nearly-inferrable Text-Hypothesis pairs where the interpretation needs reasoning. RTE3 has introduced texts of paragraph length: in turn this has prompted us to upgrade VENSES by the addition of a discourse level anaphora resolution module, which is paramount to allow entailment in pairs where the relevant portion of text contains pronominal expressions. We present the system, its relevance to the task at hand and an evaluation.

A grammatical component for a text-to-speech system

A grammatical component to supply information to a text-to-speech system is presented. It is composed of four modules: a lexicon, a morphological recognizer, a syntactic preanalyzer, a parser. The parser is composed of a bottom-up algorithm implementing a context-free grammar for Italian simply as Recursive Transition Networks (RTN). No conditions are introduced on the arcs: this will constitute the topic of future work. The paper deals extensively with Focontours and focus assignment rules in Italian by introducing Pierrehumberts notational system [4]. We found that an Fopeak usually obtains in coincidence with the focussed syllable: this may be either realized as a fall or as a rise followed or preceded by a trailing/leading Fomovement in the opposite direction. The combination of the two tones High/Low and Low/High can appear also on other constituents - Phonological Words - but the steepness of the jump or dipping is remarkably inferior to the one occurring in focussed constituents.

Text understanding with GETARUNS for Q/A and summarization

Summarization and Question Answering need precise linguistic information with a much higher coverage than what is being offered by currently available statistically based systems. We assume that the starting point of any interesting application in these fields must necessarily be a good syntacticsemantic parser. In this paper we present the system for text understanding called GETARUNS, General Text and Reference Understanding System (Delmonte, 2003a). The heart of the system is a rule-based top-down DCG-style parser, which uses an LFG oriented grammar organization. The parser produces an f-structure as a DAG which is then used to create a Logical Form, the basis for all further semantic representation. GETARUNS, has a highly sophisticated linguistically based semantic module which is used to build up the Discourse Model. Semantic processing is strongly modularized and distributed amongst a number of different submodules which take care of Spatio-Temporal Reasoning, Discourse Level Anaphora Resolution.

SPARSAR: An Expressive Poetry Reader

We present SPARSAR, a system for the automatic analysis of poetry(and text) style which makes use of NLP tools like tokenizers, sentence splitters, NER (Name Entity Recognition) tools, and taggers. In addition the system adds syntactic and semantic structural analysis and prosodic modeling. We do a dependency mapping to analyse the verbal complex and determine Discourse Structure. Another important component of the system is a phonological parser to account for OOVWs, in the process of grapheme to phoneme conversion of the poem. We also measure the prosody of the poem by associating mean durational values in msecs to each syllable from a database of syllable durations; to account for missing syllables we built a syllable parser with the aim to evaluate durational values for any possible syllable structure. A fundamental component for the production of emotions is the one that performs affective and sentiment analysis. This is done on a line by line basis. Lines associated to specific emotions are then marked to be pronounced with special care for the final module of the system, which is reponsible for the production of expressive reading by a TTS module, in our case the one made available by Apple on their computers. Expressive reading is allowed by the possibility to interact with the TTS.

Evaluating GETARUNS parser with GREVAL test suite

GREVAL, the test suite of 500 English sentences taken from SUSANNE Corpus and made available by John Carroll and Ted Briscoe at their website, has been used to test the performance of a symbolic linguistically-based parser called GETARUNS presented in (Delmonte, 2002). GETARUNS is a symbolic linguistically-based parser written in Prolog Horn clauses which uses a strong deterministic policy by means of a lookahead mechanism and a WFST. The grammar allows the specification of linguistic rules in a highly declarative mode: it works topdown and by making a heavy use of linguistic knowledge may achieve an almost complete deterministic policy: in this sense it is equivalent to an LR parser. The results obtained fare higher that the ones reported in (Preis, 2003) and this we argue is due basically to the symbolic rule-based approach: we reach 96% precision (coverage) and 84% recall (accuracy). We assume that from a psycholinguistic point of view, parsing requires setting up a number of disambiguating strategies, to tell arguments apart from adjuncts and reduce the effects of backtracking. To do that the system is based on LFG theoretical framework and uses Grammatical Functions information to help the parser cope with syntactic ambiguity. In the paper we shall comment on some shortcomings of the Greval corpus annotation and more in general we shall criticize some aspects of the Dependency Structure representation.

Feedback generation and linguistic knowledge in 'SLIM' automatic tutor

SLIM is a prototype interactive multimedia self-learning linguistic software for foreign language students at beginner-false beginner level. It allows students to work both in an autonomous self-directed mode or in a way of programmed learning in which the process of self-instruction is pre-programmed and monitored. In this latter mode it incorporates assessment and evaluation tools in order to behave as an automatic tutor. It is organized into three basic components: audiovisual materials; a linguistic database recording all language material in text format; the supervisor. Audiovisual materials are partially taken from commercially available courses; the linguistic database is a highly sophisticated classification of all words and utterances of the course, both in written and spoken form, from all possible linguistic aspects. The supervisor is both an attractive, enjoyable and strongly pedagogically based software that allows the user to work on language materials. The most outstanding feature of SLIM is the use of speech analysis and recognition which is a fundamental aspect of all second language learning programmes. We also assume that a learning model can be represented by a finite state automaton made up by a fixed number of possible states – corresponding to the macro and microlevels at which the students competence may be modelled – each one being internally constituted by the actual linguistic objects of knowledge of the language that make it up.