This paper investigates an isolated setting of the lexical substitution task of replacing words with their synonyms. In particular, we examine this problem in the setting of subtitle generation and evaluate state of the art scoring methods that predict the validity of a given substitution. The paper evaluates two context independent models and two contextual models. The major findings suggest t...

We present a system to determine content similarity of documents. Our goal is to identify pairs of book chapters that are translations of the same original chapter. Achieving this goal requires identification of not only the different topics in the documents but also of the particular flow of these topics. Our approach to content similarity evaluation employs ngrams of lexical chains and measur...

Exploiting Linguistic Knowledge in Lexical and Compositional Semantic Models Tong Wang Doctor of Philosophy Graduate Department of Computer Science University of Toronto 2016 A fundamental principle in distributional semantic models is to use similarity in linguistic environment as a proxy for similarity in meaning. Known as the distributional hypothesis, the principle has been successfully app...

This paper describes our submission for the *SEM shared task of Semantic Textual Similarity. We estimate the semantic similarity between two sentences using regression models with features: 1) n-gram hit rates (lexical matches) between sentences, 2) lexical semantic similarity between non-matching words, 3) string similarity metrics, 4) affective content similarity and 5) sentence length. Domai...

Graphs have become ubiquitous structures to encode geographic knowledge online. The Semantic Web’s linked open data, folksonomies, wiki websites and open gazetteers can be seen as geo-knowledge graphs, that is labeled graphs whose vertices represent geographic concepts and whose edges encode the relations between concepts. To compute the semantic similarity of concepts in such structures, this ...

Even in languages that do not share script, bilinguals process cognates faster than matched noncognates in a range of tasks. The current research more fully explores what underpins the cognate 'advantage' in different script bilinguals (Japanese-English). To do this, instead of the more traditional binary cognate/noncognate distinction, the current study uses continuous measures of phonological...

Ten experiments were conducted on visually presented Serbo-Croatian words and pseudowords, comprising phonemically similar and dissimilar context-target sequences. There were five main results. First, phonemic similarity effects in both lexical decision and naming are independent of graphemic similarity. Second, phonemic similarity need not facilitate lexical decision; the direction of its effe...

Four experiments investigated acoustic-phonetic similarity in the mapping process between the speech signal and lexical representations (vertical similarity). Auditory stimuli were used where ambiguous initial phonemes rendered a phoneme sequence lexically ambiguous (perceptual-lexical ambiguities). A cross-modal priming paradigm (Experiments 1, 2, and 3) showed facilitation for targets related...