Few-shot relation extraction constitutes a critical task in natural language processing. Its aim is to train model using limited number of labeled samples when data are scarce, thereby enabling the rapidly learn and accurately identify relationships between entities within textual data. Prototypical networks extensively utilized for simplicity efficiency few-shot scenarios. Nevertheless, prototypical derive their prototypes by averaging feature instances given category. In cases where...

Linguistic metonymy is a common type of figurative language in natural processing (NLP), where concept represented by closely associated word or phrase, for example "business executives suits". As result, resolution has become an important NLP task aimed at correctly identifying metonymic expressions within sentences. Previous approaches to this have typically relied on pre-trained models (PLMs) using fine-tuning process. However, can be time-consuming and resource-intensive, may lead loss...

Unmanned surface vehicles (USVs) have been widely used in research and exploration, patrol, defense. Autonomous navigation obstacle avoidance, as the essential technology of USVs, are key conditions for successful mission execution. However, fine modeling conventional algorithms cannot meet real-time precise behavior control strategy USVs complex environments, which poses a great challenge to autonomous policy. In this paper, deep reinforcement learning-based UANOA (USVs avoidance) method is...

Automatic navigation with collision‐free has become a critical challenge for unmanned surface vehicles (USVs) to expand their application scenarios. Conventional methods achieving automatic of USVs typically rely on finely modeling the environment, thus exhibiting poor generalization capabilities. Methods based deep reinforcement learning possess powerful abilities and have achieved promising results in USV‐automatic navigation‐tasks. However, increase complexity network model structures led...

Summary A corpus (eg, patents or news texts) is an important knowledge resource that contains various topics, such as specific technologies social events. Topic detection models of corpus, eg, Latent Dirichlet Allocation and KeyGraph, provide basis for exploring the status quo trends in science, technology, However, these suffer from low retrieval performance they only consider text own explicit semantics a single‐domain corpus. In addition, many incremental models, online‐LDA, are based on...

Summary Taxonomic relations play an important role in various Natural Language Processing (NLP) tasks (eg, information extraction, question answering and knowledge inference). Existing approaches on embedding‐based taxonomic relation learning mainly rely the word embeddings trained using co‐occurrence‐based similarity learning. However, performance of these is not quite satisfactory due to lack sufficient semantic within embeddings. To solve this problem, we propose improved approach learn...

Abstract Entity synonyms play a significant role in entity-based tasks. Previous approaches use linguistic syntax, distributional, and semantic features to expand entity synonym sets from text corpora. Due the flexibility complexity of Chinese language expression, aforementioned are still difficult robustly text, because these fail track holistic semantics among entities suffer error propagation. This paper introduces an approach for expanding based on bilateral context filtering strategy....

The accuracy of autonomous navigation and obstacle avoidance unmanned aerial vehicles (UAVs) in complex environments has become one challenging task. In this paper, an the UAV (ANOAU) algorithm based on deep reinforcement learning (DRL) been proposed to achieve accurate path planning environments. our work, we use actor–critic‐based DRL framework control from sensor input output UAV’s action design a set reward functions that can be adapted for environment. Meanwhile, alleviate...

In this paper, we propose a mobile robot autonomous navigation method based on deep recurrent Q-network, which realizes the end-to-end mapping of robot's perception its behavior. study, first extract features separately from own state information and sensor's detection surrounding environment to form current real-time observation. Then, concatenated is input into neural network, representation moment enhanced by combining historical overcome problem bias in evaluation under incomplete...

Abstract Given a question and its answer candidates (named QA corpus), selection is the task of identifying most relevant answers to question. Answer widely used in answering, web search, so on. Current deep neural network models primarily utilize local features extracted from input question‐answer pairs (QA pairs). However, global contained corpora are under‐utilized, we argue that these substantially contribute task. To verify this point view, propose novel model combines for selection. In...

<title>Abstract</title> Class imbalance inevitably occurs in dynamic data stream scenarios and can pose tremendous challenges for mining. To address these challenges, an adaptive resampling weighted ensemble method (ARWE) is proposed this paper. First, the subdivision Poisson (DSPR) module ARWE developed to class problem thedata stream. DSPR combines local information from minority samples with rate design a sample-weighting scheme that enhance visibility of samples, particularly those at...

Automatically generating entity synonym sets (i.e., of terms that represent the same entity) is an important work for many entity-based tasks. Existing studies on set generation either use a ranking plus pruning approach or take problem as two-phase task extracting synonymy pairs, subsequently organizing these pairs into sets). However, approaches ignore association semantics entities and suffer from error propagation issue. In this paper, we propose neural-network-based exploits information...

Entity synonyms play an important role in natural language processing applications, such as query expansion and question answering. There are three main distribution characteristics web texts:1) appearing parallel structures; 2) occurring with specific patterns sentences; 3) distributed similar contexts. The first second rely on reliable prior knowledge susceptive to data sparseness, bringing high accuracy low recall synonym extraction. third one may lead but accuracy, since it identifies a...

Synonyms play an important role in many entity-based applications. However, most known synonym extraction methods are English, while Chinese ones relatively rare. In this paper, we propose a simple yet effective and cleaning framework that automatically extracts entity synonyms from encyclopedias. the phase, three named direct extraction, pattern-based neural mining employed to acquire large number of candidate sets multiple sources online encyclopedias (e.g., title, infobox abstract). error...

Entity synonym sets play a significant role in many entity-based tasks. Robustly discovering entity requires analyzing the linguistic and semantic features of entities target language. Although there are approaches to expand from English text, it is still difficult robustly Chinese text because flexibility complexity language expression. In this paper, we propose an approach for expanding via bilateral context filtering strategies. Specifically, mainly includes three parts. First, gains...