The Experts below are selected from a list of 32559 Experts worldwide ranked by ideXlab platform
Amir Shoham - One of the best experts on this subject based on the ideXlab platform.
-
does mother tongue make for women s work linguistics household labor and gender identity
Journal of Economic Behavior and Organization, 2015Co-Authors: Daniel L Hicks, Estefania Santacreuvasut, Amir ShohamAbstract:This paper studies the formation and persistence of gender identity in a sample of U.S. immigrants. We show that gender roles are acquired early in life, and once established, persist regardless of how long an individual has lived in the U.S. We use a novel approach relying on linguistic variation and document that households with individuals whose native language emphasizes gender in its Grammatical Structure are significantly more likely to allocate household tasks on the basis of sex and to do so more intensively. We present evidence of two mechanisms for our observed associations – that languages serve as cultural markers for origin country norms or that features of language directly influence cognition and behavior. Our findings do not appear to be driven by plausible alternatives such as selection in migration and marriage markets, as gender norms of behavior are evident even in the behavior of single person households.
-
does mother tongue make for women s work linguistics household labor and gender identity
Social Science Research Network, 2014Co-Authors: Daniel L Hicks, Estefania Santacreuvasut, Amir ShohamAbstract:This paper studies the formation and transmission of gender identity in a sample of U.S. immigrants. We document that households with individuals whose native language emphasizes gender in its Grammatical Structure are significantly more likely to allocate household tasks on the basis of sex and to do so more intensively. These gender identities are evident even in the behavior of single person households. To isolate the role of language, we employ a differences-in-differences analysis based on the critical period hypothesis of language acquisition. This analysis demonstrates that time allocations are particularly skewed only for immigrants who arrive after previously learning a gender marked native language. Once established, gender norms persist regardless of how long an individual has lived in the U.S. However, we find little evidence of intergenerational transmission of these identities, and instead document rapid cultural assimilation among second-generation immigrants.
Daniel L Hicks - One of the best experts on this subject based on the ideXlab platform.
-
does mother tongue make for women s work linguistics household labor and gender identity
Journal of Economic Behavior and Organization, 2015Co-Authors: Daniel L Hicks, Estefania Santacreuvasut, Amir ShohamAbstract:This paper studies the formation and persistence of gender identity in a sample of U.S. immigrants. We show that gender roles are acquired early in life, and once established, persist regardless of how long an individual has lived in the U.S. We use a novel approach relying on linguistic variation and document that households with individuals whose native language emphasizes gender in its Grammatical Structure are significantly more likely to allocate household tasks on the basis of sex and to do so more intensively. We present evidence of two mechanisms for our observed associations – that languages serve as cultural markers for origin country norms or that features of language directly influence cognition and behavior. Our findings do not appear to be driven by plausible alternatives such as selection in migration and marriage markets, as gender norms of behavior are evident even in the behavior of single person households.
-
does mother tongue make for women s work linguistics household labor and gender identity
Social Science Research Network, 2014Co-Authors: Daniel L Hicks, Estefania Santacreuvasut, Amir ShohamAbstract:This paper studies the formation and transmission of gender identity in a sample of U.S. immigrants. We document that households with individuals whose native language emphasizes gender in its Grammatical Structure are significantly more likely to allocate household tasks on the basis of sex and to do so more intensively. These gender identities are evident even in the behavior of single person households. To isolate the role of language, we employ a differences-in-differences analysis based on the critical period hypothesis of language acquisition. This analysis demonstrates that time allocations are particularly skewed only for immigrants who arrive after previously learning a gender marked native language. Once established, gender norms persist regardless of how long an individual has lived in the U.S. However, we find little evidence of intergenerational transmission of these identities, and instead document rapid cultural assimilation among second-generation immigrants.
Pascale Fung - One of the best experts on this subject based on the ideXlab platform.
-
code switched language models using neural based synthetic data from parallel sentences
Conference on Computational Natural Language Learning, 2019Co-Authors: Genta Indra Winata, Andrea Madotto, Pascale FungAbstract:Training code-switched language models is difficult due to lack of data and complexity in the Grammatical Structure. Linguistic constraint theories have been used for decades to generate artificial code-switching sentences to cope with this issue. However, this require external word alignments or constituency parsers that create erroneous results on distant languages. We propose a sequence-to-sequence model using a copy mechanism to generate code-switching data by leveraging parallel monolingual translations from a limited source of code-switching data. The model learns how to combine words from parallel sentences and identifies when to switch one language to the other. Moreover, it captures code-switching constraints by attending and aligning the words in inputs, without requiring any external knowledge. Based on experimental results, the language model trained with the generated sentences achieves state-of-the-art performance and improves end-to-end automatic speech recognition.
Mehrnoosh Sadrzadeh - One of the best experts on this subject based on the ideXlab platform.
-
Quantization, Frobenius and Bi Algebras from the Categorical Framework of Quantum Mechanics to Natural Language Semantics
Frontiers of Physics in China, 2017Co-Authors: Mehrnoosh SadrzadehAbstract:Compact Closed categories and Frobenius and Bi algebras have been applied to model and reason about Quantum protocols. The same constructions have also been applied to reason about natural language semantics under the name: ``categorical distributional compositional'' semantics, or in short, the ``DisCoCat'' model. This model combines the statistical vector models of word meaning with the compositional models of Grammatical Structure. It has been applied to natural language tasks such as disambiguation, paraphrasing and entailment of phrases and sentences. The passage from the Grammatical Structure to vectors is provided by a functor, similar to the Quantization functor of Quantum Field Theory. The original DisCoCat model only used compact closed categories. Later, Frobenius algebras were added to it to model long distance dependancies such as relative pronouns. Recently, bialgebras have been added to the pack to reason about quantifiers. This paper reviews these constructions and their application to natural language semantics. We go over the theory and present some of the core experimental results.
-
a quantum teleportation inspired algorithm produces sentence meaning from word meaning and Grammatical Structure 15
arXiv: Computation and Language, 2014Co-Authors: Twareque S Ali, Mehrnoosh Sadrzadeh, S R Clark, Bob Coecke, Edward Grefenstette, S Hasibul, Hassan Chowdhury, Kazuo Fujikawa, Saeid MolladavoudiAbstract:We discuss an algorithm which produces the meaning of a sentence given meanings of its words, and its resemblance to quantum teleportation. In fact, this protocol was the main source of inspiration for this algorithm which has many applications in the area of Natural Language Processing.
-
a quantum teleportation inspired algorithm produces sentence meaning from word meaning and Grammatical Structure
Malaysian Journal of Mathematical Sciences, 2013Co-Authors: S R Clark, Bob Coecke, Edward Grefenstette, Stephen Pulman, Mehrnoosh SadrzadehAbstract:We discuss an algorithm which produces the meaning of a sentence given meanings of its words, and its resemblance to quantum teleportation. In fact, this protocol was the main source of inspiration for this algorithm which has many applications in the area of Natural Language Processing. Quantum teleportation (Bennett et al., 1993) is one of the most conceptually challenging and practically useful concepts that has emerged from the quantum information revolution. For example, via logic-gate teleportation (Gottesman and Chuang, 1999) it gave rise to the measurement-based computational model, it also plays a key role in current investigations into the nature of quantum correlations, e.g. Skrzypczyk et al. (2009), and it even has been proposed as a model for time travel (Bennett and Schumacher, 2002). It also formed the cornerstone for a new axiomatic approach and diagrammatic calculus for quantum theory (Abramsky and Coecke, 2004; Coecke, 2010; Coecke and Duncan, 2011). i i “coecke-mjms-apcwqis” — 2014/7/21 — 13:52 — page 16 — #2 i i i i i i Stephen Clark, Bob Coecke, Edward Grefenstette, Stephen Pulman and Mehrnoosh Sadrzadeh Arguably, when such a radically new concept emerges in a novel foundational area of scientific investigation, one may expect that the resulting conceptual and structural insights could also lead to progress in other areas, something which has happened on many occasions in the history of physics. In the context of quantum information, for example, it is well-known that quantum complexity theory has helped solve many problems in classical complexity theory. Here we explain how a high-level description of quantum teleportation with emphasis on information flows has successfully helped solve a longstanding open problem in the area of Natural Language Processing (NLP), and the problem of modeling meaning for natural language more generally (Clark et al., 2008; Coecke et al., 2010). This work featured as a cover heading in the New Scientist (11 Dec. 2011) (Aron, 2010), and has been experimentally tested for its capability to perform key NLP tasks such as word sense disambiguation in context (Grefenstette and Sadrzadeh, 2011).1 1. The NLP Problem Dictionaries explain the meanings of words; however, in natural language words are organized as sentences, but we don’t have dictionaries that explain the meanings of sentences. Still, a sentence carries more information than the words it is made up from; e.g. meaning(Alice sends a message to Bob) 6= meaning(Bob sends a message to Alice). Evidently, this is where Grammatical Structure comes into play. Consequently, we as humans must use some algorithm that converts the meanings of words, via the Grammatical Structure, into the meaning of a sentence. All of this may seem to be only of academic interest; however, search engines such as Google face exactly the same challenge. They typically read a string of words as a ‘bag of words’, ignoring the Grammatical Structure. This is simply because (until recently) there was no mathematical model for assigning meanings to sentences.2 On the other hand, there is a widely used model for word meaning, the vector space model (Schutze, 1998). This vector space model of word meaning works as follows. One chooses a set of context words which will form the basis vectors of a vector space.3 Given a word to which one wishes to assign meaning,e.g. ‘Alice’, one relies on 1EMNLP is the leading conference on corpus-based experimental NLP. 2More precisely, there was no mathematical model for assigning meanings to sentences that went beyond truthfulness. Montague semantics (Thomason, 1974) is a compositional model of meaning, but at most assigns truth values to sentences, and evidently there is more to sentence meaning than the mere binary assignment of either true or false. 3These context words may include nouns, verbs etc.; the vector space model built from the British National Corpus typically contains 10s of thousands of these words as basis vectors. 16 Malaysian Journal of Mathematical Sciences i i “coecke-mjms-apcwqis” — 2014/7/21 — 13:52 — page 17 — #3 i i i i i i A quantum teleportation inspired algorithm produces sentence meaning from word meaning and Grammatical Structure a large corpus, e.g. (part of) the web, to establish the relative frequency that ‘Alice’ occurs ‘close’ to each of these basis words. The list of all these relative frequencies yields a vector that represents this word, its meaning vector. Now, if one wants to verify synonymy of two words, it suffices to compute the innerproduct of the meaning vectors of these words, and verify how close it is to 1. Indeed, since synonyms are interchangeable, one would expect them to typically occur in the context of the same words, and hence their meaning vectors should be the same in the statistical limit. For example, in a corpus mainly consisting of computer science literature, one would expect Alice and Bob to always occur in the same context and hence their meaning vectors would almost be the same. Of course, if the corpus were English literature (cf. Carroll (1865)), then this similarity would break down. Until recently, the state of affairs in computational linguistics was one of two separate communities (Gazdar, 1996). One community focused on noncompositional purely distributional methods such as the vector space model described above. The other community studied the compositional mathematical Structure of sentences, building on work by Chomsky (1957), Lambek (1958) and Montague (Thomason, 1974). This work is mainly about the Grammatical Structure of sentences; Grammatical type calculi are algebraic gadgets that allow one to verify whether a sentence has a correct Grammatical Structure. 2. Caps, cups, and teleportation In Abramsky and Coecke (2004), a novel axiomatic framework was proposed to reason about quantum informatic processes, which admits a sound and faithful purely diagrammatic calculus (Coecke, 2010); for some more recent developments we refer to Coecke and Duncan (2011). Ideal post-selected teleportation provides the cornerstone for the diagrammatic reasoning techniques, e.g. here is the derivation of the general teleportation protocol where the f -label represents both the measurement outcome and the corresponding correction performed by Bob Coecke (2010):
-
mathematical foundations for a compositional distributional model of meaning
Lambek Festschirft‚ Linguistic Analysis‚ vol. 36, 2010Co-Authors: Bob Coecke, Mehrnoosh Sadrzadeh, Stephen ClarkAbstract:We propose a mathematical framework for a unification of the distributional theory of meaning in terms of vector space models, and a compositional theory for Grammatical types, for which we rely on the algebra of Pregroups, introduced by Lambek. This mathematical framework enables us to compute the meaning of a well-typed sentence from the meanings of its constituents. Concretely, the type reductions of Pregroups are ‘lifted’ to morphisms in a category, a procedure that transforms meanings of constituents into a meaning of the (well-typed) whole. Importantly, meanings of whole sentences live in a single space, independent of the Grammatical Structure of the sentence. Hence the inner-product can be used to compare meanings of arbitrary sentences, as it is for comparing the meanings of words in the distributional model. The mathematical Structure we employ admits a purely diagrammatic calculus which exposes how the information flows between the words in a sentence in order to make up the meaning of the whole sentence. A variation of our ‘categorical model’ which involves constraining the scalars of the vector spaces to the semiring of Booleans results in a Montague-style Boolean-valued semantics.
Mark Craven - One of the best experts on this subject based on the ideXlab platform.
-
representing sentence Structure in hidden markov models for information extraction
International Joint Conference on Artificial Intelligence, 2001Co-Authors: Soumya Ray, Mark CravenAbstract:We study the application of Hidden Markov Models (HMMs) to learning information extractors for -ary relations from free text. We propose an approach to representing the Grammatical Structure of sentences in the states of the model. We also investigate using an objective function during HMM training which maximizes the ability of the learned models to identify the phrases of interest. We evaluate our methods by deriving extractors for two binary relations in biomedical domains. Our experiments indicate that our approach learns more accurate models than several baseline approaches.
