Search

The work presented in this paper reviews in depth the computational methods and tecniques developed within a project aiming to annotate CORIS/CODIS with part-of-speech (PoS) tags. In a large number of studies devoted to automatic PoS annotation the tagsets tend to be pre-defined and, consequently, theory oriented. Our aim is to automatically derive an empirically founded PoS classification making few a priori assumptions about the Pos classes to be distinguished. Early approaches to this problem were based on the hypothesis that if two words are syntactically and semantically different, they will appear in different contexts. There are a number of studies based on this hypothesis in the fields of both computational linguistics and cognitive science aiming at building automatic or semi-automatic procedures for clustering. These papers examine the distributional behaviour of target words by comparing the lexical distribution of their respective collocates and by using quantitative measures of distributional similarity. The main drawback of these techniques is the limited context of analysis. Information is collected from a restricted context of, for instance, 3 words which can conceal syntactic dependencies longer than the context interval. Our approach to solving this problem is to use basic syntactic relations together with distributional information. The algorithm extracts information from loosely labelled dependency structure that encode only basic and broadly accepted syntactic relations, namely Head/Dependent and the distinction of dependents into Argument vs. Adjunct. Such information is exploited to further refine an exclusively distributional classification induced by means of Brill’s algorithm. Three main uncontroversial classes emerge from this broad range process: noun, verbs and all the others. Syntactic information extracted from the dependency structures is automatically processed and encoded in formulae (or syntactic types) lexically anchored. The algorithm proposed creates pairs of words and syntactic types and connects each pair in accordance with syntactic similarities between them, producing an extensive graph. The algorithm then exploits statistical information from the graph so far obtained in order to achieve a first level breakdown of PoS classes. The ultimate PoS tagset is obtained as a further decomposition of the previous phases by using distributional knowledge and a more sophisticated clustering metric. In order to evaluate the effectiveness of the proposed PoS tagsets a number of experiments have been carried out: the results obtained using state of the art tagging experiments will be presented.

The present paper focuses on the first results of a work in progress about grammar induction for Italian. This work aims at achieving a better understanding of the models behind Italian syntax and at building systems able to automatically create Italian Treebanks. The experiments of automatic induction for lexical categories, that was the focus of a FIRB project, are our starting points to induce complex syntactic types which will be then used to semi-automatically build Categorial Grammar (CG) derivations of a given corpus. The study carried so far is based on a specific version of CG, namely Categorial Type Logic (CTL, the logic version of CG, Moortgat ‘97). An important aspect of this system is the derivability relation that holds between syntactic categories and allows to avoid multiple assignments of lexical types when unnecessary. The starting point of our type induction algorithm (Buszkowski and Penn ‘90) must be a Treebank consisting of binary trees. In particular, so far we have based our preliminary work on the Turin University Treebank (TUT). The latter consists of dependency grammar trees for 1500 sentences that we have re-written into binary trees so to extract categorial types by exploiting the dependency relations of the original trees. In this contribution we present the set of categories induced from TUT and illustrate how we intend to exploit CTL derivability relation to simplify the lexicon and avoid unnecessary multiple assignments. In particular, we plan to make use of both syntactic and semantic filtering criteria as well as statistical clustering methods. Due to the richness of dependency and semantic information encoded into categorial grammar types, once we have induced a lexicon as described above, it will be possible to run a syntactic analyser (parser) based on CTL in order to study language diversities between Italian, English and Dutch, since for the latter two languages similar grammar induction works have been already carried out. Finally, a last aspect on which we will bring the attention is the need of enhancing a CTL parser’s performance by taking into account semantic information so to discard semantically implausible derivation. Connected to this aim is the need of extending the lexical assignments with semantic meaning representation (see Bos ‘05). It’s plausible to think that the integration of CG with semantic resources like FrameNet could help achieving this ultimate goal.

The DiaCORIS project aims at the construction of a diachronic corpus comprising written Italian texts produced between 1861 and 1945, extending the structure and the research possibilities of the synchronic 100-million word corpus CORIS/CODIS. A preliminary in depth study has been performed in order to design a representative and well balanced sample of the Italian language over a time period that contains all the main events of contemporary Italian history from the National Unification to the end of the Second World War. The paper describes in detail such design processes as the definition of the main subcorpora and their proportions, the type of documents inserted in each part of the corpus, the document annotation schema and the technological infrastructure designed to manage the corpus access as well as the web interface to corpus data.

We aim to automatically induce a PoS tagset for Italian by analysing the distributional behaviour of Italian words. To this end, we propose an algorithm that (a) extracts information from loosely labelled dependency structures that encode only basic and broadly accepted syntactic relations, namely Head/Dependent and the distinction of dependents into Argument vs. Adjunct, and (b) derives a possible set of word classes. The paper reports on some preliminary experiments carried out using the induced tagset in conjunction with state-of-the-art PoS taggers. The method proposed to design a proper tagset exploits little, if any, language-specific knowledge: hence it is in principle applicable to any language.