SGT-Net: A Transformer-Based Stratified Graph Convolutional Network for 3D Point Cloud Semantic Segmentation

In recent years,semantic segmentation on 3D point cloud data has attracted much attention.Unlike 2D images where pixels distribute regularly in the image domain,3D point clouds in non-Euclidean space are irregular and inherently sparse.Therefore,it is very difficult to extract long-range contexts and effectively aggregate local features for semantic segmentation in 3D point cloud space.Most current methods either focus on local feature aggregation or long-range context dependency,but fail to directly establish a global-local feature extractor to complete the point cloud semantic segmentation tasks.In this paper,we propose a Transformer-based stratified graph convolutional network(SGT-Net),which enlarges the effective receptive field and builds direct long-range dependency.Specifically,we first propose a novel dense-sparse sampling strategy that provides dense local vertices and sparse long-distance vertices for subsequent graph convolutional network(GCN).Secondly,we propose a multi-key self-attention mechanism based on the Transformer to further weight augmentation for crucial neighboring relationships and enlarge the effective receptive field.In addition,to further improve the efficiency of the network,we propose a similarity measurement module to determine whether the neighborhood near the center point is effective.We demonstrate the validity and superiority of our method on the S3DIS and ShapeNet datasets.Through ablation experiments and segmentation visualization,we verify that the SGT model can improve the performance of the point cloud semantic segmentation.

ST-SIGMA:Spatio-temporal semantics and interaction graph aggregation for multi-agent perception and trajectory forecasting

Scene perception and trajectory forecasting are two fundamental challenges that are crucial to a safe and reliable autonomous driving(AD)system.However,most proposed methods aim at addressing one of the two challenges mentioned above with a single model.To tackle this dilemma,this paper proposes spatio-temporal semantics and interaction graph aggregation for multi-agent perception and trajectory forecasting(STSIGMA),an efficient end-to-end method to jointly and accurately perceive the AD environment and forecast the trajectories of the surrounding traffic agents within a unified framework.ST-SIGMA adopts a trident encoder-decoder architecture to learn scene semantics and agent interaction information on bird’s-eye view(BEV)maps simultaneously.Specifically,an iterative aggregation network is first employed as the scene semantic encoder(SSE)to learn diverse scene information.To preserve dynamic interactions of traffic agents,ST-SIGMA further exploits a spatio-temporal graph network as the graph interaction encoder.Meanwhile,a simple yet efficient feature fusion method to fuse semantic and interaction features into a unified feature space as the input to a novel hierarchical aggregation decoder for downstream prediction tasks is designed.Extensive experiments on the nuScenes data set have demonstrated that the proposed ST-SIGMA achieves significant improvements compared to the state-of-theart(SOTA)methods in terms of scene perception and trajectory forecasting,respectively.Therefore,the proposed approach outperforms SOTA in terms of model generalisation and robustness and is therefore more feasible for deployment in realworld AD scenarios.

SHEL:a semantically enhanced hardware-friendly entity linking method

With the help of pre-trained language models,the accuracy of the entity linking task has made great strides in recent years.However,most models with excellent performance require fine-tuning on a large amount of training data using large pre-trained language models,which is a hardware threshold to accomplish this task.Some researchers have achieved competitive results with less training data through ingenious methods,such as utilizing information provided by the named entity recognition model.This paper presents a novel semantic-enhancement-based entity linking approach,named semantically enhanced hardware-friendly entity linking(SHEL),which is designed to be hardware friendly and efficient while maintaining good performance.Specifically,SHEL's semantic enhancement approach consists of three aspects:(1)semantic compression of entity descriptions using a text summarization model;(2)maximizing the capture of mention contexts using asymmetric heuristics;(3)calculating a fixed size mention representation through pooling operations.These series of semantic enhancement methods effectively improve the model's ability to capture semantic information while taking into account the hardware constraints,and significantly improve the model's convergence speed by more than 50%compared with the strong baseline model proposed in this paper.In terms of performance,SHEL is comparable to the previous method,with superior performance on six well-established datasets,even though SHEL is trained using a smaller pre-trained language model as the encoder.

Learning Semantic Lexicons Using Graph Mutual Reinforcement Based Bootstrapping

这份报纸论述一个方法基于图用一个新引导方法学习语义词典相互的加强(GMR ) 。途径使用仅仅未标记的数据和一些种子词为每个语义范畴学习新词。与另外的引导方法不同，我们使用基于 GMR 的引导排序候选人词和模式。试验性的结果证明基于 GMR 的引导途径在在里面域数据和外面域数据两个都超过存在算法。而且，它证明结果取决于语料库而且质量的尺寸不仅。

Introducing semantic information into motion graph

To improve motion graph based motion synthesis,semantic control was introduced.Hybrid motion features including both numerical and user-defined semantic relational features were extracted to encode the characteristic aspects contained in the character's poses of the given motion sequences.Motion templates were then automatically derived from the training motions for capturing the spatio-temporal characteristics of an entire given class of semantically related motions.The data streams of motion documents were automatically annotated with semantic motion class labels by matching their respective motion class templates.Finally,the semantic control was introduced into motion graph based human motion synthesis.Experiments of motion synthesis demonstrate the effectiveness of the approach which enables users higher level of semantically intuitive control and high quality in human motion synthesis from motion capture database.

Semantic-aware graph convolution network on multi-hop paths for link prediction

Knowledge graph(KG) link prediction aims to address the problem of missing multiple valid triples in KGs. Existing approaches either struggle to efficiently model the message passing process of multi-hop paths or lack transparency of model prediction principles. In this paper,a new graph convolutional network path semantic-aware graph convolution network(PSGCN) is proposed to achieve modeling the semantic information of multi-hop paths. PSGCN first uses a random walk strategy to obtain all-hop paths in KGs,then captures the semantics of the paths by Word2Sec and long shortterm memory(LSTM) models,and finally converts them into a potential representation for the graph convolution network(GCN) messaging process. PSGCN combines path-based inference methods and graph neural networks to achieve better interpretability and scalability. In addition,to ensure the robustness of the model,the value of the path thresholdKis experimented on the FB15K-237 and WN18RR datasets,and the final results prove the effectiveness of the model.

Typos Correction in Overseas Chinese Learning Based on Chinese Character Semantic Knowledge Graph

In recent years, more and more foreigners begin to learn Chinese characters, but they often make typos when using Chinese. The fundamental reason is that they mainly learn Chinese characters from the glyph and pronunciation, but do not master the semantics of Chinese characters. If they can understand the meaning of Chinese characters and form knowledge groups of the characters with relevant meanings, it can effectively improve learning efficiency. We achieve this goal by building a Chinese character semantic knowledge graph (CCSKG). In the process of building the knowledge graph, the semantic computing capacity of HowNet was utilized, and 104,187 associated edges were finally established for 6752 Chinese characters. Thanks to the development of deep learning, OpenHowNet releases the core data of HowNet and provides useful APIs for calculating the similarity between two words based on sememes. Therefore our method combines the advantages of data-driven and knowledge-driven. The proposed method treats Chinese sentences as subgraphs of the CCSKG and uses graph algorithms to correct Chinese typos and achieve good results. The experimental results show that compared with keras-bert and pycorrector + ernie, our method reduces the false acceptance rate by 38.28% and improves the recall rate by 40.91% in the field of learning Chinese as a foreign language. The CCSKG can help to promote Chinese overseas communication and international education.

Semantic Link Network Based Knowledge Graph Representation and Construction

A knowledge graph consists of a set of interconnected typed entities and their attributes,which shows a better performance to organize,manage and understand knowledge.However,because knowledge graphs contain a lot of knowledge triples,it is difficult to directly display to researchers.Semantic Link Network is an attempt,and it can deal with the construction,representation and reasoning of semantics naturally.Based on the Semantic Link Network,this paper explores the representation and construction of knowledge graph,and develops an academic knowledge graph prototype system to realize the representation,construction and visualization of knowledge graph.

New method for query answering in semantic web

To promote the efficiency of knowledge base retrieval based on description logic, the concept of assertional graph （AG）, which is directed labeled graph, is defined and a new AG-based retrieval method is put forward. This method converts the knowledge base and query clause into knowledge AG and query AG by making use of the given rules and then makes use of graph traversal to carry out knowledge base retrieval. The experiment indicates that the efficiency of this method exceeds, respectively, the popular RACER and KAON2 system by 0.4% and 3.3%. This method can obviously promote the efficiency of knowledge base retrieval.

Annotation and Retrieval System of CAD Models Based on Functional Semantics

CAD model retrieval based on functional semantics is more significant than content-based 3D model retrieval during the mechanical conceptual design phase. However, relevant research is still not fully discussed. Therefore, a functional semantic-based CAD model annotation and retrieval method is proposed to support mechanical conceptual design and design reuse, inspire designer creativity through existing CAD models, shorten design cycle, and reduce costs. Firstly, the CAD model functional semantic ontology is constructed to formally represent the functional semantics of CAD models and describe the mechanical conceptual design space comprehensively and consistently. Secondly, an approach to represent CAD models as attributed adjacency graphs（AAG） is proposed. In this method, the geometry and topology data are extracted from STEP models. On the basis of AAG, the functional semantics of CAD models are annotated semi-automatically by matching CAD models that contain the partial features of which functional semantics have been annotated manually, thereby constructing CAD Model Repository that supports model retrieval based on functional semantics. Thirdly, a CAD model retrieval algorithm that supports multi-function extended retrieval is proposed to explore more potential creative design knowledge in the semantic level. Finally, a prototype system, called Functional Semantic-based CAD Model Annotation and Retrieval System（FSMARS）, is implemented. A case demonstrates that FSMARS can successfully botain multiple potential CAD models that conform to the desired function. The proposed research addresses actual needs and presents a new way to acquire CAD models in the mechanical conceptual design phase.

Construction and Resource Locating of Semantic P2P Grid Based on Description Logics

This paper proposes an algorithm applied in se mantic P2P network based on the description logics with the purpose for realizing the concepts distribution of resources, which makes the resources semantic locating easy. With the idea of the consistent hashing in the Chord, our algorithm stores the addresses and resources with the values of the same type to select instance. In addition, each peer has its own ontology, which will be completed by the knowledge distributed over the network during the exchange of CHGs （classification hierarchy graphs）. The hierarchy classification of concepts allows to find matching resource by querying to the upper level concept because the all concepts described in the CHG have the same root.

Adapting Property Path for Polynomial-Time Evaluation and Reasoning on Semantic Web

Property path is the latest navigational extension of the standard query language SPARQL 1.1 for the Semantic Web.However,in the existing SPARQL query systems which support property path,the query efficiency is very low and does not support reasoning.This paper proposes a new existential semantics which has polynomial-time evaluation complexity and an equivalent relationship with the current semantics,and transforms the property path expressions to the extended nested regular expressions based on the existential semantics and proves the semantic equivalence after the transformation considering the RDFS semantics.The property path query engine is achieved by implementing the nested regular expressions algorithm and the transformation rules from the property path expressions to the nested regular expressions,which maintains the syntax simplicity of property path and the goal-oriented polynomial-time reasoning to avoid computing the RDF graph closure.The experiment results not only show the characteristics of query engine based on the existential semantics in efficiency and reasoning,but also further validate the equivalence between the results based on current semantics and those based on the existential semantics for property path after the removal of duplicate values.

Novel Epistemic and Predictive Heuristic for Semantic and Dynamic Social Networks Analysis

Using Kripke semantics, we have identified and reduced an epistemic incompleteness in the metaphor commonly employed in Social Networks Analysis (SNA), which basically compares information flows with current flows in advanced centrality measures. Our theoretical approach defines a new paradigm for the semantic and dynamic analysis of social networks including shared content. Based on our theoretical findings, we define a semantic and predictive model of dynamic SNA for Enterprises Social Networks (ESN), and experiment it on a real dataset.

Machine Learning Meets the Semantic Web

Remarkable progress in research has shown the efficiency of Knowledge Graphs(KGs)in extracting valuable external knowledge in various domains.A Knowledge Graph(KG)can illustrate high-order relations that connect two objects with one or multiple related attributes.The emerging Graph Neural Networks(GNN)can extract both object characteristics and relations from KGs.This paper presents how Machine Learning(ML)meets the Semantic Web and how KGs are related to Neural Networks and Deep Learning.The paper also highlights important aspects of this area of research,discussing open issues such as the bias hidden in KGs at different levels of graph representation。

Explanatory Multi-Scale Adversarial Semantic Embedding Space Learning for Zero-Shot Recognition

The goal of zero-shot recognition is to classify classes it has never seen before, which needs to build a bridge between seen and unseen classes through semantic embedding space. Therefore, semantic embedding space learning plays an important role in zero-shot recognition. Among existing works, semantic embedding space is mainly taken by user-defined attribute vectors. However, the discriminative information included in the user-defined attribute vector is limited. In this paper, we propose to learn an extra latent attribute space automatically to produce a more generalized and discriminative semantic embedded space. To prevent the bias problem, both user-defined attribute vector and latent attribute space are optimized by adversarial learning with auto-encoders. We also propose to reconstruct semantic patterns produced by explanatory graphs, which can make semantic embedding space more sensitive to usefully semantic information and less sensitive to useless information. The proposed method is evaluated on the AwA2 and CUB dataset. These results show that our proposed method achieves superior performance.

A Survey of Knowledge Graph Construction Using Machine Learning

Knowledge graph(KG)serves as a specialized semantic network that encapsulates intricate relationships among real-world entities within a structured framework.This framework facilitates a transformation in information retrieval,transitioning it from mere string matching to far more sophisticated entity matching.In this transformative process,the advancement of artificial intelligence and intelligent information services is invigorated.Meanwhile,the role ofmachine learningmethod in the construction of KG is important,and these techniques have already achieved initial success.This article embarks on a comprehensive journey through the last strides in the field of KG via machine learning.With a profound amalgamation of cutting-edge research in machine learning,this article undertakes a systematical exploration of KG construction methods in three distinct phases:entity learning,ontology learning,and knowledge reasoning.Especially,a meticulous dissection of machine learningdriven algorithms is conducted,spotlighting their contributions to critical facets such as entity extraction,relation extraction,entity linking,and link prediction.Moreover,this article also provides an analysis of the unresolved challenges and emerging trajectories that beckon within the expansive application of machine learning-fueled,large-scale KG construction.

Identifying Semantic in High-Dimensional Web Data Using Latent Semantic Manifold

Latent Semantic Analysis involves natural language processing techniques for analyzing relationships between a set of documents and the terms they contain, by producing a set of concepts (related to the documents and terms) called semantic topics. These semantic topics assist search engine users by providing leads to the more relevant document. We develope a novel algorithm called Latent Semantic Manifold (LSM) that can identify the semantic topics in the high-dimensional web data. The LSM algorithm is established upon the concepts of topology and probability. Asearch tool is also developed using the LSM algorithm. This search tool is deployed for two years at two sites in Taiwan: 1) Taipei Medical University Library, Taipei, and 2) Biomedical Engineering Laboratory, Institute of Biomedical Engineering, National Taiwan University, Taipei. We evaluate the effectiveness and efficiency of the LSM algorithm by comparing with other contemporary algorithms. The results show that the LSM algorithm outperforms compared with others. This algorithm can be used to enhance the functionality of currently available search engines.

基于图卷积神经网络的三维点云分割算法Graph⁃PointNet

三维点云无序不规则的特性使得传统的卷积神经网络无法直接应用,且大多数点云深度学习模型往往忽略大量的空间信息。为便于捕获空间点邻域信息,获得更好的点云分析性能以用于点云语义分割,文中提出Graph⁃PointNet点云深度学习模型。Graph⁃PointNet在经典点云模型PointNet的基础上,结合二维图像中聚类思想,设计了图卷积特征提取模块取代多层感知器嵌入PointNet中。图卷积特征提取模块首先通过K近邻算法搜寻相邻特征点组成图结构,接着将多组图结构送入图卷积神经网络提取局部特征用于分割。同时文中设计一种新型点云采样方法多邻域采样,多邻域采样通过设置点云间夹角阈值,将点云区分为特征区域和非特征区域,特征区域用于提取特征,非特征区域用于消除噪声。对室内场景S3DIS、室外场景Semantic3D数据集进行实验,得到二者整体精度分别达到89.33%和89.78%,平均交并比达到64.62%,61.47%,均达到最佳效果。最后,进行消融实验,进一步证明了文中所提出的多邻域采样和图卷积特征提取模块对提高点云语义分割的有效性。

Graph reasoning over explicit semantic relation

Multi-hop reasoning over language or graphs represents a significant challenge in contemporary research,particularly with the reliance on deep neural networks.These networks are integral to text reasoning processes,yet they present challenges in extracting and representing domain or commonsense knowledge,and they often lack robust logical reasoning capabilities.To address these issues,we introduce an innovative text reasoning framework.This framework is grounded in the use of a semantic relation graph and a graph neural network,designed to enhance the model’s ability to encapsulate knowledge and facilitate complex multi-hop reasoning.Our framework operates by extracting knowledge from a broad range of texts.It constructs a semantic relationship graph based on the logical relationships inherent in the reasoning process.Beginning with the core question,the framework methodically deduces key knowledge,using it as a guide to iteratively establish a complete evidence chain,thereby determining the final answer.Leveraging the advanced reasoning capabilities of the graph neural network,this approach is adept at multi-hop logical reasoning.It demonstrates strong performance in tasks like machine reading comprehension and question answering,while also clearly delineating the path of logical reasoning.

Extracting Semantic Subgraphs to Capture the Real Meanings of Ontology Elements

An element may have heterogeneous semantic interpretations in different ontologies. Therefore, understanding the real local meanings of elements is very useful for ontology operations such as querying and reasoning, which are the foundations for many applications including semantic searching, ontology matching, and linked data analysis. However, since different ontologies have different preferences to describe their elements, obtaining the semantic context of an element is an open problem. A semantic subgraph was proposed to capture the real meanings of ontology elements. To extract the semantic subgraphs, a hybrid ontology graph is used to represent the semantic relations between elements. An extracting algorithm based on an electrical circuit model is then used with new conductivity calculation rules to improve the quality of the semantic subgraphs. The evaluation results show that the semantic subgraphs properly capture the local meanings of elements. Ontology matching based on semantic subgraphs also demonstrates that the semantic subgraph is a promising technique for ontology applications.