Page: 1 2 3 4 5 6 7... 328
42 |
Comparative Quality Estimation for Machine Translation. An Application of Artificial Intelligence on Language Technology using Machine Learning of Human Preferences ...
|
|
|
|
BASE
|
|
Show details
|
|
45 |
Sentiment polarity shifters : creating lexical resources through manual annotation and bootstrapped machine learning ...
|
|
|
|
BASE
|
|
Show details
|
|
46 |
Induktive Kategorienbildung in der Inhaltsanalyse: Kombination automatischer und manueller Verfahren
|
|
|
|
In: Forum Qualitative Sozialforschung / Forum: Qualitative Social Research ; 20 ; 1 ; 30 (2019)
|
|
BASE
|
|
Show details
|
|
47 |
Project Description - hermA: Automated modelling of hermeneutic processes
|
|
|
|
In: Hamburger Journal für Kulturanthropologie ; 7 ; 119-123 ; {Digital Heritage} (2019)
|
|
BASE
|
|
Show details
|
|
48 |
Untersuchungen zur Methodik und Effizienz der tastaturbasierten Eingabeverfahren verschiedener Schriftsysteme der Welt
|
|
Wang, Kai. - : Justus-Liebig-Universität Gießen, 2019. : FB 05 - Sprache, Literatur, Kultur. Germanistik, 2019
|
|
BASE
|
|
Show details
|
|
49 |
Sentiment polarity shifters : creating lexical resources through manual annotation and bootstrapped machine learning
|
|
Schulder, Marc. - : Saarländische Universitäts- und Landesbibliothek, 2019
|
|
BASE
|
|
Show details
|
|
50 |
Comparative Quality Estimation for Machine Translation. An Application of Artificial Intelligence on Language Technology using Machine Learning of Human Preferences
|
|
|
|
BASE
|
|
Show details
|
|
51 |
The cooccurrence of linguistic structures ; Die Kookkurrenz sprachlicher Strukturen
|
|
|
|
BASE
|
|
Show details
|
|
52 |
Automatic Analysis of Linguistic Complexity and Its Application in Language Learning Research
|
|
|
|
BASE
|
|
Show details
|
|
53 |
Composition Models for the Representation and Semantic Interpretation of Nominal Compounds
|
|
|
|
Abstract:
The central topic of this thesis are composition models of distributional semantics and their application for representing the semantics of German and English nominal compounds. Composition models are mathematical transformations that, given a compound like Apfelbaum ‘apple tree’, can be applied to the vector representations of Apfel ‘apple’ and Baum ‘tree’ to obtain a vector representation for the compound Apfelbaum ‘apple tree’. The new composed representation is deemed appropriate if it is similar to the representation of Apfelbaum that can be directly learned from large corpora using distributional methods. The thesis is structured into eight chapters. The first four chapters introduce compounds from a linguistic perspective (Chapter 1), present a review of annotation schemes for nominal compounds and introduce a new hybrid annotation scheme (Chapter 2), introduce neural networks and how to represent words via numerical features (Chapter 3) and detail the distributional representation of words (Chapter 4). Existing composition models of distributional semantics are reviewed and evaluated in Chapter 5. Chapter 5 also introduces three new composition models: addmask, wmask and multimatrix, that aim to improve over existing composition models either though a more efficient parametrization (*mask) or by promoting parameter reuse across different, but semantically similar words (multimatrix). The results show that composition models are able to construct meaningful composed representations for 81.8% of the German test compounds, and 78.03% of the English test compounds. In Chapter 6 composed representations are shown to be a useful indicator when investigating non-compositional (lexicalized) compounds. For example, when modeling a compound like Tigerauge, ‘tiger eye’, composition models will produce a composed representation that corresponds to the literal interpretation of the compound - the eye of a tiger. This vector, however, is dissimilar to the distributional vector learned directly from the corpus which captures the lexicalized meaning of semi-precious stone. In Chapter 7 composed representations prove to be the best features for classifying compounds in terms of their semantic relations in setups where simplex words and compounds have representations of the same length. Further analyses show also that some of the modifier information is discarded during the composition process and that extrinsic evaluations tasks such as the semantic classification task are necessary for assessing and improving the quality of the composed representations. Chapter 8 concludes by emphasizing the main contributions of the thesis and sketching directions for future contributions.
|
|
Keyword:
004; 400; 420; 430; Composition Models; Computational Linguistics; Computerlinguistik; Distributional Semantics; Kompositum; Machine Learning; Maschinelles Lernen; Neural Networks; Nominal Compounds; Semantic Relations; Semantik; Word Representations
|
|
URL: http://nbn-resolving.de/urn:nbn:de:bsz:21-dspace-870984 http://hdl.handle.net/10900/87098 https://doi.org/10.15496/publikation-28485
|
|
BASE
|
|
Hide details
|
|
Page: 1 2 3 4 5 6 7... 328
|
|