Candidate Entities Research Articles

A segment enhanced span-based model for nested named entity recognition

Named entity recognition (NER) is a fundamental problem in natural language processing. In particular, nested entities are commonly existed in real-life textual data for the NER task. However, the current span-based methods for nested NER are computationally expensive, lacking of explicit boundary supervision and generating many negative samples for span classification, which affect their overall performance. In this paper, we propose a Segment Enhanced Span-based model for nested NER (SESNER). The proposed model treats the nested NER task as a segment covering problem. First, it models entities as segments, detects the segment endpoints and identifies the positional relationship between neighboring endpoints. Then, it detects the outermost segments to generate candidate entity spans nested in it for span classification. Our proposed model has the advantages of enhancing boundary supervision in learning span representations by detecting segment endpoints, reducing the number of negative samples without losing long entities that are ignored by most span-based methods, and improving runtime performance. Moreover, a novel augmented training mechanism is also proposed to further improve the model performance by extending the training dataset with data that were wrongly predicted before. Experimental results show that our proposed SESNER model has achieved promising performance with near linear time complexity on the benchmark datasets.

Analysis of Big Data Embedded Mapping Model Oriented to Prediction of Subordinate Relationship

With the rapid development of information technology, research on big data information has gone deep into all walks of life, and massive amounts of data spew out anytime and anywhere. This paper proposes a domain-subordinate relationship acquisition that combines multiple strategies and word representations. From semi-structured text and unstructured text, the candidate entity upper-lower relationship entity pairs are extracted, and then the obtained candidate upper-lower relationship entity pairs are verified by the support vector machine method to obtain high-quality candidate entity pairs. In the training phase, an exploratory method is designed to search for the optimal solution, and a greedy mechanism is introduced to evaluate the effectiveness of the reinforcement learning agent, so that the virtual node mapping scheme can continuously move towards higher system returns. Evolution has finally achieved the optimal virtual network mapping decision, so that large-scale tasks can be deployed to task processing nodes in the appropriate underlying network to achieve efficient task execution in a big data environment. The experimental results show that the Tard method can effectively avoid model overfitting and improve the task recognition processing ability in the actual application process under the premise of meeting large-scale task requests.

A Semantic-Embedding Model-Driven Seq2Seq Method for Domain-Oriented Entity Linking on Resource-Restricted Devices

General entity linking systems usually leverage global coherence of all the mapped entities in the same document by using semantic embeddings and graph-based approaches. However, graph-based approaches are computationally expensive for open-domain datasets. In this paper, the authors overcome these problems by presenting an RDF embedding-based seq2seq entity linking method in specific domains. They filter candidate entities of mentions having similar meanings by using the domain information of the annotated pairs. They resolve high ambiguous pairs by using Bi-directional long short-term memory (Bi-LSTM) and attention mechanism for the entity disambiguation. To evaluate the system with baseline methods, they generate a dataset including book, music, and movie categories. They achieved 0.55 (Mi-F1), 0.586 (Ma-F1), 0.846 (Mi-F1), and 0.87 (Ma-F1) scores for high and low ambiguous datasets. They compare the method by using recent (WNED-CWEB) datasets with existing methods. Considering the domain-specificity of the proposed method, it tends to achieve competitive results while using the domain-oriented datasets.

Representation Learning with LDA Models for Entity Disambiguation in Specific Domains

Entity disambiguation is extremely important in knowledge construction. The word representation model ignores the influence of the ordering between words on the sentence or text information. Thus, we propose a domain entity disambiguation method that fuses the doc2vec and LDA topic models. In this study, the doc2vec document is used to indicate that the model obtains the vector form of the entity reference item and the candidate entity from the domain corpus and knowledge base, respectively. Moreover, the context similarity and category referential similarity calculations are performed based on the knowledge base of the upper and lower relation domains that are constructed. The LDA topic model and doc2vec model are used to obtain word expressions with different meanings of polysemic words. We use the k-means algorithm to cluster the word vectors under different topics to obtain the topic domain keywords of the text, and perform the similarity calculations under the domain keywords of the different topics. Finally, the similarities of the three feature types are merged and the candidate entity with the highest similarity degree is used as the final target entity. The experimental results demonstrate that the proposed method outperforms the existing model, which proves its feasibility and effectiveness.

Knowledge-aware Named Entity Recognition with Alleviating Heterogeneity

Named Entity Recognition (NER) is a fundamental and important research topic for many downstream NLP tasks, aiming at detecting and classifying named entities (NEs) mentioned in unstructured text into pre-defined categories. Learning from labeled data only is far from enough when it comes to domain-specific or temporally-evolving entities (medical terminologies or restaurant names). Luckily, open-source Knowledge Bases (KBs) (Wikidata and Freebase) contain NEs that are manually labeled with predefined types in different domains, which is potentially beneficial to identify entity boundaries and recognize entity types more accurately. However, the type system of a domain-specific NER task is typically independent of that of current KBs and thus exhibits heterogeneity issue inevitably, which makes matching between the original NER and KB types (Person in NER potentially matches President in KBs) less likely, or introduces unintended noises without considering domain-specific knowledge (Band in NER should be mapped to Out_of_Entity_Types in the restaurant-related task). To better incorporate and denoise the abundant knowledge in KBs, we propose a new KB-aware NER framework (KaNa), which utilizes type-heterogeneous knowledge to improve NER. Specifically, for an entity mention along with a set of candidate entities that are linked from KBs, KaNa first uses a type projection mechanism that maps the mention type and entity types into a shared space to homogenize the heterogeneous entity types. Then, based on projected types, a noise detector filters out certain less-confident candidate entities in an unsupervised manner. Finally, the filtered mention-entity pairs are injected into a NER model as a graph to predict answers. The experimental results demonstrate KaNa's state-of-the-art performance on five public benchmark datasets from different domains.

Read, Retrospect, Select: An MRC Framework to Short Text Entity Linking

Entity linking (EL) for the rapidly growing short text (e.g. search queries and news titles) is critical to industrial applications. Most existing approaches relying on adequate context for long text EL are not effective for the concise and sparse short text. In this paper, we propose a novel framework called Multi-turn Multiple-choice Machine reading comprehension (M3) to solve the short text EL from a new perspective: a query is generated for each ambiguous mention exploiting its surrounding context, and an option selection module is employed to identify the golden entity from candidates using the query. In this way, M3 framework sufficiently interacts limited context with candidate entities during the encoding process, as well as implicitly considers the dissimilarities inside the candidate bunch in the selection stage. In addition, we design a two-stage verifier incorporated into M3 to address the commonly existed unlinkable problem in short text. To further consider the topical coherence and interdependence among referred entities, M3 leverages a multi-turn fashion to deal with mentions in a sequence manner by retrospecting historical cues. Evaluation shows that our M3 framework achieves the state-of-the-art performance on five Chinese and English datasets for the real-world short text EL.

TPmod: A Tendency-Guided Prediction Model for Temporal Knowledge Graph Completion

Temporal knowledge graphs (TKGs) have become useful resources for numerous Artificial Intelligence applications, but they are far from completeness. Inferring missing events in temporal knowledge graphs is a fundamental and challenging task. However, most existing methods solely focus on entity features or consider the entities and relations in a disjoint manner. They do not integrate the features of entities and relations in their modeling process. In this paper, we propose TPmod, a tendency-guided prediction model, to predict the missing events for TKGs (extrapolation). Differing from existing works, we propose two definitions for TKGs: the Goodness of relations and the Closeness of entity pairs. More importantly, inspired by the attention mechanism, we propose a novel tendency strategy to guide our aggregated process. It integrates the features of entities and relations, and assigns varying weights to different past events. What is more, we select the Gate Recurrent Unit (GRU) as our sequential encoder to model the temporal dependency in TKGs. Besides, the Softmax function is employed to generate the final decreasing group of candidate entities. We evaluate our model on two TKG datasets: GDELT-5 and ICEWS-250. Experimental results show that our method has a significant and consistent improvement compared to state-of-the-art baselines.

Identifying comparable entities from online question-answering contents

As an emerging platform, online question-answering (Q&A) communities are becoming valuable corpus sources that reflect the opinions of expert consumers on comparable entities. In this study, a novel method-Identifying Comparable entities from online Question-Answering contents (ICQA)-is proposed to effectively extract comparable entities from online Q&A communities. In ICQA, candidate entities are firstly extracted by utilizing the advantages of pattern-based methods and supervised learning-based methods. An entity comparison network is then built by considering the credibility difference and entity relatedness in Q&A contents to analyze competitiveness between entities and extract comparable entities from candidate entities accordingly. Thus, within the same method framework, comparable entities and their competitiveness ranks can be simultaneously identified for a given entity specified by managers or consumers. By taking the automotive industry as the experimental background, the effectiveness of ICQA is demonstrated with comprehensive experiments regarding comparable entity identification and comparable entity ranking. Experimental results demonstrate that compared with state-of-the-art methods, ICQA can identify more accurate and broader comparable entities, find novel entities ignored by other methods, and provide a more solid comparable entity rank-features that are deemed desirable and applicable for both managers and consumers in making rational decisions based on online Q&A contents.

Cross-Modal Progressive Comprehension for Referring Segmentation.

Given a natural language expression and an image/video, the goal of referring segmentation is to produce the pixel-level masks of the entities described by the subject of the expression. Previous approaches tackle this problem by implicit feature interaction and fusion between visual and linguistic modalities in a one-stage manner. However, human tends to solve the referring problem in a progressive manner based on informative words in the expression, i.e., first roughly locating candidate entities and then distinguishing the target one. In this paper, we propose a cross-modal progressive comprehension (CMPC) scheme to effectively mimic human behaviors and implement it as a CMPC-I (Image) module and a CMPC-V (Video) module to improve referring image and video segmentation models. For image data, our CMPC-I module first employs entity and attribute words to perceive all the related entities that might be considered by the expression. Then, the relational words are adopted to highlight the target entity as well as suppress other irrelevant ones by spatial graph reasoning. For video data, our CMPC-V module further exploits action words based on CMPC-I to highlight the correct entity matched with the action cues by temporal graph reasoning. In addition to the CMPC, we also introduce a simple yet effective Text-Guided Feature Exchange (TGFE) module to integrate the reasoned multimodal features corresponding to different levels in the visual backbone under the guidance of textual information. In this way, multi-level features can communicate with each other and be mutually refined based on the textual context. Combining CMPC-I or CMPC-V with TGFE can form our image or video version referring segmentation frameworks and our frameworks achieve new state-of-the-art performances on four referring image segmentation benchmarks and three referring video segmentation benchmarks respectively. Our code is available at https://github.com/spyflying/CMPC-Refseg.

A Knowledge Graph Entity Disambiguation Method Based on Entity-Relationship Embedding and Graph Structure Embedding.

The purpose of knowledge graph entity disambiguation is to match the ambiguous entities to the corresponding entities in the knowledge graph. Current entity ambiguity elimination methods usually use the context information of the entity and its attributes to obtain the mention embedding vector, compare it with the candidate entity embedding vector for similarity, and perform entity matching through the similarity. The disadvantage of this type of method is that it ignores the structural characteristics of the knowledge graph where the entity is located, that is, the connection between the entity and the entity, and therefore cannot obtain the global semantic features of the entity. To improve the Precision and Recall of entity disambiguation problems, we propose the EDEGE (Entity Disambiguation based on Entity and Graph Embedding) method, which utilizes the semantic embedding vector of entity relationship and the embedding vector of subgraph structure feature. EDEGE first trains the semantic vector of the entity relationship, then trains the graph structure vector of the subgraph where the entity is located, and balances the weights of these two vectors through the entity similarity function. Finally, the balanced vector is input into the graph neural network, and the matching between the entities is output to achieve entity disambiguation. Extensive experimental results proved the effectiveness of the proposed method. Among them, on the ACE2004 data set, the Precision, Recall, and F1 values of EDEGE are 9.2%, 7%, and 11.2% higher than baseline methods.

Entity Alignment Method of Points of Interest for Internet Location-Based Services

In Internet applications, the description for the same point of interest (POI) entity for different location-based services (LBSs) is not completely identical. The POI entity information in a single LBS data source contains incomplete data and exhibits insufficient objectivity. Aligning and consolidating POI entities from various LBSs can provide users with more comprehensive, objective, and authoritative POI information. We herein propose a multi-attribute measurement-based entity alignment method for Internet LBSs to achieve POI entity alignment and data consolidation. This method is based on multi-attribute information (geographical information, text coincidence information, semantic information) of POI entities and is combined with different measurement methods to calculate the similarity of candidate entity pairs. Considering the demand for computational efficiency, the particle swarm optimization algorithm is used to train the model and optimize the weights of multi-attribute measurements. A consolidation strategy is designed for the LBS text data and user rating data from different sources to obtain more comprehensive and objective information. The experimental results show that, compared with other baseline models, the POI alignment method based on multi-attribute measurement performed the best. Using this method, the information of POI entities in multisource LBS can be integrated to serve netizens.

Identifying comparable entities with indirectly associative relations and word embeddings from web search logs

Comparable entity identification plays an essential role in the decision making of both consumers and firms in competitive environment. In contrast to traditional cooccurrence approaches, this paper proposes a novel method, namely, ICE (identifying comparable entities) for effectively identifying comparable entities from web search logs, which are online user-generated contents that reflect users' attention and preferences. ICE consists of two stages: the formulation of directly and indirectly associative relations, followed by a generative procedure that is designed for deriving a broad set of candidate entities that are indirectly associative with a specified focal entity; and a deep-learning-based semantic analysis with a word embedding procedure for measuring the similarities between entity profiles so as to target comparable entities from the candidate set. Extensive experiments show that ICE outperforms several baseline methods in the identification of accurate, broad and novel comparable entities with suitable rankings.

Improving Candidate Generation for Low-resource Cross-lingual Entity Linking

Cross-lingual entity linking (XEL) is the task of finding referents in a target-language knowledge base (KB) for mentions extracted from source-language texts. The first step of (X)EL is candidate generation, which retrieves a list of plausible candidate entities from the target-language KB for each mention. Approaches based on resources from Wikipedia have proven successful in the realm of relatively high-resource languages, but these do not extend well to low-resource languages with few, if any, Wikipedia pages. Recently, transfer learning methods have been shown to reduce the demand for resources in the low-resource languages by utilizing resources in closely related languages, but the performance still lags far behind their high-resource counterparts. In this paper, we first assess the problems faced by current entity candidate generation methods for low-resource XEL, then propose three improvements that (1) reduce the disconnect between entity mentions and KB entries, and (2) improve the robustness of the model to low-resource scenarios. The methods are simple, but effective: We experiment with our approach on seven XEL datasets and find that they yield an average gain of 16.9% in Top-30 gold candidate recall, compared with state-of-the-art baselines. Our improved model also yields an average gain of 7.9% in in-KB accuracy of end-to-end XEL. 1

Recalibration convolutional networks for learning interaction knowledge graph embedding

Knowledge graph embedding aims to learn the embedded representation of entities and relations in knowledge graphs which is very important for the subsequent link prediction task. However, two key issues are existed for learning knowledge graph embedding: 1) How to take full advantage of the deep learning algorithms to generate expressive embeddings? 2) How to solve the polysemy phenomenon caused by multi-relations knowledge graphs that entities and relations show different semantics after involving different predictions? In this article, to tackle the first problem, the multi-layer convolutional networks are adopted to generate features about entities and relations then used to predict candidate entity. Moreover, the representation power of the networks is strengthened by integrating an effective recalibration mechanism which can accentuate informative features selectively. To tackle the second problem, we propose to learn multiple specific interaction embeddings. Instead of directly learning one general embedding to preserve all information for each entity and relation, their interactions are captured to model the cross-semantic influence from relations to entities and from entities to relations. Compared to traditional embedding models, the proposed model can provide more generalization capabilities and effectively capture potential links between entities and relations. Experimental results have revealed that the proposed model achieves the state-of-the-art performance for general evaluation metrics on link prediction tasks.

Named entity translation method based on machine translation lexicon

In the context of the rapid development of computer technology, communication between various languages has become increasingly important. Among the research methods of named entities, the research on named entity translation methods based on machine translation is one of the research hotspots. Named entity translation is to realize the switching between entities in two languages, which can be used by browsers, translators, etc., and can greatly reduce the cost of communication between people from all over the world. Due to the immaturity of the existing translation model technology and the lack of integration, the translation of bilingual named entities with unique composition is very challenging. Based on this, this paper proposes a fusion method of bilingual entity class named entity translation based on chunk symmetry strategy and English–Chinese transliteration model based on machine learning strategy. According to the bilingual corpus, a more standard candidate entity translation pair is generated through bilingual named entity alignment. The transliteration model is used to reorder and correct the candidate translation results, so as to achieve the correct selection of translation pairs. Experiments show that the model based on the translation of named entity translations based on chunks and transliteration models based on machine learning strategies not only effectively solves the problem of difficulty in word ordering and selection in bilingual translation, but also makes extraction to a certain extent. The accuracy of the translation results has been improved.

A span-graph neural model for overlapping entity relation extraction in biomedical texts.

Entity relation extraction is one of the fundamental tasks in biomedical text mining, which is usually solved by the models from natural language processing. Compared with traditional pipeline methods, joint methods can avoid the error propagation from entity to relation, giving better performances. However, the existing joint models are built upon sequential scheme, and fail to detect overlapping entity and relation, which are ubiquitous in biomedical texts. The main reason is that sequential models have relatively weaker power in capturing long-range dependencies, which results in lower performance in encoding longer sentences. In this article, we propose a novel span-graph neural model for jointly extracting overlapping entity relation in biomedical texts. Our model treats the task as relation triplets prediction, and builds the entity-graph by enumerating possible candidate entity spans. The proposed model captures the relationship between the correlated entities via a span scorer and a relation scorer, respectively, and finally outputs all valid relational triplets. Experimental results on two biomedical entity relation extraction tasks, including drug-drug interaction detection and protein-protein interaction detection, show that the proposed method outperforms previous models by a substantial margin, demonstrating the effectiveness of span-graph-based method for overlapping relation extraction in biomedical texts. Further in-depth analysis proves that our model is more effective in capturing the long-range dependencies for relation extraction compared with the sequential models. Related codes are made publicly available at http://github.com/Baxelyne/SpanBioER.

LoG: a locally-global model for entity disambiguation

Entity disambiguation (ED) aims to link textual mentions in a document to the correct named entities in a knowledge base (KB). Although global ED models usually outperform local models by collectively linking mentions based on the topical coherence assumption, they may still incur incorrect entity assignment when a document contains multiple topics. Therefore, we propose a Locally-Global model (LoG) for ED which extracts global features locally, i.e., among a limited number of neighboring mentions, to combine the respective superiority of both models. In particular, we derive mention neighbors according to the syntactic distance on a dependency parse tree, and propose a tree connection method CoSimTC to measure the cross-tree distance between mentions. We also recognize the importance of keywords in a document for collective entity disambiguation, which reveal the central topic information of the document. Hence, we propose a keyword extraction method Sent2Word to detect keywords of each document. Furthermore, we extend the Graph Attention Network (GAT) to integrate both local and global features to produce a discriminative representation for each candidate entity. Our experimental results on six widely-adopted public datasets demonstrate better performance compared with state-of-the-art ED approaches. The high efficiency of the LoG model further verifies its feasibility in practice.

Entity set expansion in knowledge graph: a heterogeneous information network perspective

Entity set expansion (ESE) aims to expand an entity seed set to obtain more entities which have common properties. ESE is important for many applications such as dictionary construction and query suggestion. Traditional ESE methods relied heavily on the text and Web information of entities. Recently, some ESE methods employed knowledge graphs (KGs) to extend entities. However, they failed to effectively and efficiently utilize the rich semantics contained in a KG and ignored the text information of entities in Wikipedia. In this paper, we model a KG as a heterogeneous information network (HIN) containing multiple types of objects and relations. Fine-grained multi-type meta paths are proposed to capture the hidden relation among seed entities in a KG and thus to retrieve candidate entities. Then we rank the entities according to the meta path based structural similarity. Furthermore, to utilize the text description of entities in Wikipedia, we propose an extended model CoMeSE++ which combines both structural information revealed by a KG and text information in Wikipedia for ESE. Extensive experiments on real-world datasets demonstrate that our model achieves better performance by combining structural and textual information of entities.

CoGCN: Combining co‐attention with graph convolutional network for entity linking with knowledge graphs

AbstractEntity linking is a fundamental task in natural language processing. The task of entity linking with knowledge graphs aims at linking mentions in text to their correct entities in a knowledge graph like DBpedia or YAGO2. Most of existing methods rely on hand‐designed features to model the contexts of mentions and entities, which are sparse and hard to calibrate. In this paper, we present a neural model that first combines co‐attention mechanism with graph convolutional network for entity linking with knowledge graphs, which extracts features of mentions and entities from their contexts automatically. Specifically, given the context of a mention and one of its candidate entities' context, we introduce the co‐attention mechanism to learn the relatedness between the mention context and the candidate entity context, and build the mention representation in consideration of such relatedness. Moreover, we propose a context‐aware graph convolutional network for entity representation, which takes both the graph structure of the candidate entity and its relatedness with the mention context into consideration. Experimental results show that our model consistently outperforms the baseline methods on five widely used datasets.

Enriching Context Information for Entity Linking with Web Data

Entity linking (EL) is the task of determining the identity of textual entity mentions given a predefined knowledge base (KB). Plenty of existing efforts have been made on this task using either “local” information (contextual information of the mention in the text), or “global” information (relations among candidate entities). However, either local or global information might be insufficient especially when the given text is short. To get richer local and global information for entity linking, we propose to enrich the context information for mentions by getting extra contexts from the web through web search engines (WSE). Based on the intuition above, two novel attempts are made. The first one adds web-searched results into an embedding-based method to expand the mention’s local information, where we try two different methods to help generate high-quality web contexts: one is to apply the attention mechanism and the other is to use the abstract extraction method. The second one uses the web contexts to extend the global information, i.e., finding and utilizing more extra relevant mentions from the web contexts with a graph-based model. Finally, we combine the two models we propose to use both extended local and global information from the extra web contexts. Our empirical study based on six real-world datasets shows that using extra web contexts to extend the local and the global information could effectively improve the F1 score of entity linking.