Wyniki wyszukiwania - BazTech

1

A French corpus annotated for multiword expressions and named entities

Candito Marie, Constant Mathieu, Ramisch Carlos, Savary Agata, Guillaume Bruno, Parmentier Yannick, Cordeiro Silvio Ricardo

Journal of Language Modelling

|

2020

|

Vol. 8, No. 2

415--479

EN

We present the enrichment of a French treebank of various genres with a new annotation layer for multiword expressions (MWEs) and named entities (NEs).1 Our contribution with respect to previous work on NE and MWE annotation is the particular care taken to use formal criteria, organized into decision flowcharts, shedding some light on the interactions between NEs and MWEs. Moreover, in order to cope with the well-known difficulty to draw a clear-cut frontier between compositional expressions and MWEs, we chose to use sufficient criteria only. As a result, annotated MWEs satisfy a varying number of sufficient criteria, accounting for the scalar nature of the MWE status. In addition to the span of the elements, annotation includes the subcategory of NEs (e.g., person, location) and one matching sufficient criterion for non-verbal MWEs (e.g., lexical substitution). The 3,099 sentences of the treebank were double-annotated and adjudicated, and we paid attention to cross-type consistency and compatibility with the syntactic layer. Overall inter-annotator agreement on non-verbal MWEs and NEs reached 71.1%. The released corpus contains 3,112 annotated NEs and 3,440 MWEs, and is distributed under an open license.

2

Slovak Morphosyntactic Tagset

Garabík R., Šimková M.

Journal of Language Modelling

|

2012

|

Vol. 0, No. 1

41--63

EN

Morphological annotation constitutes essential, very useful and very common linguistic information presented in corpora, especially for highly inflectional languages. The morphological tagset used in the Slovak National Corpus has been designed with several goals in mind – the tags are compact and easily human-readable, without sacrificing their informational contents. The tags consist of ASCII letters, numbers and several other characters. In general, they have a variable numer of symbols, but their order is obligatory, and each category or specific feature is assigned a particular character, which can be shared among several parts of speech. The tagset is highly functional and pragmatic, although some allowances had to be made to accommodate the traditional analysis of Slovak morphology and part of speech categories.

3

Construction of a medical corpus based on information extraction results

Marciniak M., Mykowiecka A.

Control and Cybernetics

|

2011

|

Vol. 40, no 2

337-360

EN

The paper presents a method of automatic construction of a semantically annotated corpus using the results of a rulebased information extraction (IE) application. Construction of the corpus is based on using existing programs for text tokenization and morphological analysis and combining their results with domain related correction rules. We reuse the specialized IE system to obtain a corpus annotated on the semantic level. The texts included within the corpus are Polish free text clinical data. We present the documents - diabetic patients' discharge records, the structure of the corpus annotation and the methods for obtaining the annotations. Initial evaluations based on the results of manual verification of selected data subset are also presented. The corpus, once manually corrected, is designed to be used for developing supervised machine learning models for IE applications.

4

Collecting Polish-German parallel corpora in the Web

Rosińska M.

Systems Science

|

2008

|

Vol. 34, no 4

41-45

EN

Parallel corpus has recently become an indispensable resource in multilingual natural language processing. Manual preparation of a bilingual corpus is a laborious task. Therefore methods for the automated creation of parallel corpus are currently a topic of concern for many researchers. A number of sophisticated and effective algorithms for collecting parallel texts from the Web have already been created. Unfortunately, none of them have been used in the process of Polish-German corpus creation. That is why the aim of the research has been to verify the efficiency of existing algorithms for the collecting of Polish-German parallel corpus, intended as a reference source for a Machine Translation system, to propose a new algorithm and present results achieved by the new algorithm.