RESUMEN

Entity relation extraction plays an important role in the biomedical, healthcare, and clinical research areas. Recently, pre-trained models based on transformer architectures and their variants have shown remarkable performances in various natural language processing tasks. Most of these variants were based on slight modifications in the architectural components, representation schemes and augmenting data using distant supervision methods. In distantly supervised methods, one of the main challenges is pruning out noisy samples. A similar situation can arise when the training samples are not directly available but need to be constructed from the given dataset. The BioCreative V Chemical Disease Relation (CDR) task provides a dataset that does not explicitly offer mention-level gold annotations and hence replicates the above scenario. Selecting the representative sentences from the given abstract or document text that could convey a potential entity relationship becomes essential. Most of the existing methods in literature propose to either consider the entire text or all the sentences which contain the entity mentions. This could be a computationally expensive and time consuming approach. This paper presents a novel approach to handle such scenarios, specifically in biomedical relation extraction. We propose utilizing the Shortest Dependency Path (SDP) features for constructing data samples by pruning out noisy information and selecting the most representative samples for model learning. We also utilize triplet information in model learning using the biomedical variant of BERT, viz., BioBERT. The problem is represented as a sentence pair classification task using the sentence and the entity-relation pair as input. We analyze the approach on both intra-sentential and inter-sentential relations in the CDR dataset. The proposed approach that utilizes the SDP and triplet features presents promising results, specifically on the inter-sentential relation extraction task. We make the code used for this work publicly available on Github.1.

Asunto(s)

RESUMEN

Automatic extraction of chemical-induced disease (CID) relation from unstructured text is of essential importance for disease treatment and drug development. In this task, some relational facts can only be inferred from the document rather than single sentence. Recently, researchers investigate graph-based approaches to extract relations across sentences. It iteratively combines the information from neighbor nodes to model the interactions in entity mentions that exist in different sentences. Despite their success, one severe limitation of the graph-based approaches is the over-smoothing problem, which decreases the model distinguishing ability. In this paper, we propose CID-GCN, an effective Graph Convolutional Networks (GCNs) with gating mechanism, for CID relation extraction. Specifically, we construct a heterogeneous graph which contains mention, sentence and entity nodes. Then, the graph convolution operation is employed to aggregate interactive information on the constructed graph. Particularly, we combine gating mechanism with the graph convolution operation to address the over-smoothing problem. The experimental results demonstrate that our approach significantly outperforms the baselines.

RESUMEN

The current study explored bilingual parent and child code-switching patterns over time. Concurrent and predictive models of code-switching behaviour on executive function outcomes were also examined in a sample of 29 French-English bilinguals at 36 (Wave 1) and 61 (Wave 2) months of age. We investigated whether code-switching typology in a single-language context predicted executive function performance at each wave independently, and whether growth in code-switching frequency across waves predicted executive function performance at Wave 2. At both waves, parents and children participated in two free play sessions (in English and French), followed by a battery of executive function tasks administered in the dominant language. Results indicate more frequent code-switching from the non-dominant to the dominant language in children, and that children code-switch to fill lexical gaps. Results also suggest that less frequent code-switching in a single-language context is associated with better inhibitory control skills during the preschool period.

RESUMEN

Increasing evidence suggests that language switching is a distinct form of bilingual language control that engages cognitive control. The most relevant and widely discussed framework is the Adaptive Control Hypothesis. This theoretical framework identifies language switching to be a key aspect of bilingual language control. It proposes that bilinguals' engagement in three different types of interactional contexts (single-language context, dual-language context, and dense code-switching context) confers adaptive effects on cognitive control processes. These contexts differ in the presence of both languages and how language control is exercised. The model makes predictions about behavioral outcomes associated with these contexts. This study is a novel attempt to test for the model's assumptions, predictions, and its interactional contexts. It seeks to examine the relationship between language switching behaviors, reported bilingual interactional contexts, and verbal and non-verbal cognitive control through this theoretical framework. Seventy-four English-Mandarin young adult bilinguals were measured on their self-reported engagements in the different interactional contexts and production of word and sentential language switches through experimental language switching tasks (alternating, semi-cued, and uncued switching). Cognitive control processes in verbal and non-verbal goal maintenance, interference control, selective response inhibition, and task engagement and disengagement were measured. Overall, partial support for the model was observed. Higher reported engagement in the dual-language context was positively but not uniquely related to cognitive engagement and disengagement on verbal tasks. Non-verbal goal maintenance and interference control, on the other hand, were related to uncued inter-sentential language switching. However, the distinction of the model's three interactional contexts might not be evident in a multilingual society, as findings suggest that there is fluidity in bilinguals' interactional contexts. Current findings reveal the complex interaction of language switching with distinct domains and cognitive control processes. This study is significant in testing an influential bilingual language control model.

RESUMEN

To code-switch or not to code-switch? This is a dilemma for many bilingual language teachers. In this study, the influence of teachers' CS on bilingual children's language and cognitive development is explored within heritage language (HL) classes in Singapore. Specifically, the relationship between children's language output, vocabulary development, and cognitive flexibility to teachers' classroom CS behavior, is examined within 20 preschool HL classrooms (10 Mandarin, 6 Malay, and 4 Tamil). Teachers' and children's utterances were recorded, transcribed, and analyzed for CS frequency and type (i.e., inter-sentential, intra-sentential). 173 students were assessed with receptive vocabulary and dimensional card sort tasks, and their vocabulary and cognitive switching scores assessed using correlational and mixed effects analyses. Results show that inter-sentential and intra-sentential CS frequency is positively and significantly related to children's intra-sentential CS frequency. Overall, findings revealed that teachers code-switched habitually more often than for instructional purposes. Neither inter-sentential nor intra-sentential CS was significantly related to children's development in HL vocabulary, and intra-sentential CS was found to positively and significantly relate to children's growth in cognitive flexibility. These findings reveal the multi-faceted impact of teacher's CS on children's early development.

Asunto(s)