Features For Disambiguation Research Articles

Partial multi-label learning via specific label disambiguation

Partial Multi-Label Learning (PML) aims to learn a robust multi-label classifier from training data, where each instance is associated with a set of candidate labels, among which only a subset of them is relevant. Some existing methods consider the noise in the feature space and have made some achievements. However, they ignored that each label might be only related to a subset of original features, and the other features might be noise in different label spaces. Such as, the similarity between two instances may be different in different label spaces, which is crucial in PML. To tackle the problem, we propose a novel framework named Partial multi-lAbel learning via Specific lAbel Disambiguation(PASAD), which tries to extract the label-specific features for disambiguation. Specifically, we first adapt the Hilbert–Schmidt Independence Criterion (HSIC) to identify the projection matrix for each label by maximizing the dependence between feature space and each label space. With these matrices, the instances can be mapped into each label-specific feature space to reduce irrelevant information. Besides, label correlations are considered to enrich the label-specific features. Afterward, we propose a specific label propagation method to estimate the labeling confidence values, which adapts the label propagation in each label-specific feature space and considers the interactions between different label spaces. We combine the two stages in an iterative manner. Finally, any binary classifier can be applied to induce a classification model by each label’s new specific features and credible labels. Tremendous experimental results demonstrate the effectiveness and superiority of our proposed method.

Attention Neural Network for Biomedical Word Sense Disambiguation

In order to improve the disambiguation accuracy of biomedical words, this paper proposes a disambiguation method based on the attention neural network. The biomedical word is viewed as the center. Morphology, part of speech, and semantic information from 4 adjacent lexical units are extracted as disambiguation features. The attention layer is used to generate a feature matrix. Average asymmetric convolutional neural networks (Av-ACNN) and bidirectional long short-term memory (Bi-LSTM) networks are utilized to extract features. The softmax function is applied to determine the semantic category of the biomedical word. At the same time, CNN, LSTM, and Bi-LSTM are applied to biomedical WSD. MSH corpus is adopted to optimize CNN, LSTM, Bi-LSTM, and the proposed method and testify their disambiguation performance. Experimental results show that the average disambiguation accuracy of the proposed method is improved compared with CNN, LSTM, and Bi-LSTM. The average disambiguation accuracy of the proposed method achieves 91.38%.

Biomedical Word Sense Disambiguation Based on Graph Attention Networks

Biomedical words have many semantics. Biomedical word sense disambiguation (WSD) is an important research issue in biomedicine field. Biomedical WSD refers to the process of determining meanings of ambiguous word according to its context. It is widely applied to process, translate and retrieve biomedical texts now. In order to improve WSD accuracy in biomedicine, this paper proposes a new WSD method based on graph attention neural network (GAT). Words, parts of speech, and semantic categories in context of ambiguous word are used as disambiguation features. Disambiguation features and the sentence are used as nodes to construct WSD graph. GAT is used to extract discriminative features, and softmax function is applied to determine semantic category of biomedical ambiguous word. MSH dataset is used to optimize GAT-based WSD classifier and test its accuracy. Experiments show that average accuracy of the proposed method is improved. At the same time, majority voting strategy is adopted to optimize GAT-based WSD classifier further.

3D Model Classification Based on GCN and SVM

3D model classification is an important task. Now, 3D model is usually expressed as point cloud. Disorder of point cloud brings great difficulty into 3D model classification. In order to classify 3D model correctly, a new classification method combining Graph Convolution Network (GCN) and Support Vector Machine (SVM) is proposed in this paper. Point cloud is sampled. K-Nearest Neighbor ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">KNN</i> ) algorithm is used to find K nearest points of sampling point, and adjacency matrix is established for graph convolution operation. Shape features <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D1, D2, D3</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A3</i> of sampling point are computed based on its K nearest points. Coordinates and shape features of sampling point are combined as discriminative feature. 2-layer graph convolution is used to aggregate disambiguation information of 1-degree and 2-degree adjacent points of sampling point for describing point cloud comprehensively. At the same time, maximum pooling and average pooling are adopted to retain representative information. Finally, SVM is used to classify point clouds. Experimental results show that compared with GCN based on coordinates, the proposed network improves accuracy of 3D model classification by 1.67%. Global and local information can be extracted adequately when 1024 points are sampled from point cloud. When we select 20 nearest points to compute shape features <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D1, D2, D3, A3</i> , local information of point can be described better. Shape features <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D1, D2, D3, A3</i> are combined with coordinates to describe shape and structure of point cloud better. 2-layer graph convolutions are adopted to aggregate information of 1-degree and 2-degree nodes for extracting effective disambiguation features.

Word sense disambiguation features for taxonomy extraction

Many NLP tasks, such as fact extraction, coreference reso- lution and alike, rely on existing lexical taxonomies or ontologies. One of possible ways to create a lexical taxonomy is to extract taxonomic re- lations from monolingual dictionary or encyclopedia: a semi-formalized resource designed to contain many such relations. Word-sense disam- biguation (WSD) is a mandatory tool in such approaches. Quality of extracted taxonomy greatly depends on WSD results. Most WSD approaches can be formulated as machine learning task. For this sake feature representation ranges from collocation vectors as in Lesk algorithm or neural network features in word2vec to highly specialized vector sense representation models such as AdaGram. In this paper we apply several WSD algorithms to dictionary dentions. Our main focus is on inuence of dierent approaches to extract WSD features from dictionary denitions on WSD performance.

Chinese Word Sense Disambiguation Based on Hidden Markov Model

Word sense disambiguation (WSD) is important for natural language processing. It plays important roles in information retrieval, machine translation, text categorization and topic tracking. In this paper, the transition among senses of words is considered. For an ambiguous word, its semantic codes and its left word’s semantic codes are taken as disambiguation features. At the same time, a new method based on hidden Markov model (HMM) is proposed for Chinese word sense disambiguation. Chinese Tongyici Cilin is used to determine semantic codes of words. HMM is optimized in training corpus. The WSD classifiers based on HMM is tested. Experimental results show that the accuracy of word sense disambiguation is improved.

Disambiguate Chinese Word Sense Based on Linguistics Knowledge

Word sense disambiguation (WSD) is important to many application problems in natural language processing fields, such as machine translation, parsing analysis and information retrieval. In this paper, we propose a new method to determine correct sense categories of Chinese words based on linguistics knowledge. The left word string and the right word string around the ambiguous word are respectively analyzed. Their syntactic structures are obtained for determining its intended sense. Syntactic category and part of speech are extracted as disambiguation features. A naive bayesian model is used as the classifier. Experimental results showed that the accuracy rate of classification arrives at 64%. The performance of disambiguation is improved.

Unsupervised author disambiguation using Dempster–Shafer theory

The name ambiguity problem presents many challenges for scholar finding, citation analysis and other related research fields. To attack this issue, various disambiguation methods combined with separate disambiguation features have been put forward. In this paper, we offer an unsupervised Dempster---Shafer theory (DST) based hierarchical agglomerative clustering algorithm for author disambiguation tasks. Distinct from existing methods, we exploit the DST in combination with Shannon's entropy to fuse various disambiguation features and come up with a more reliable candidate pair of clusters for amalgamation in each iteration of clustering. Also, some solutions to determine the convergence condition of the clustering process are proposed. Depending on experiments, our method outperforms three unsupervised models, and achieves comparable performances to a supervised model, while does not prescribe any hand-labelled training data.

The Interpretation of Prosody in Disjunctive Questions

Alternative questions differ prosodically from identically worded disjunctive yes/no questions in their accentual characteristics and their final pitch contour. Alternative questions are canonically pronounced with a final fall and with pitch accents on all disjuncts, while disjunctive yes/no questions are canonically pronounced with a final rise and generally without pitch accents on every disjunct. This article presents an experiment investigating the importance of these prosodic features in disambiguation. The experiment shows that the final contour is the most informative prosodic feature. Accentual characteristics also play a significant role, but, contrary to what is often assumed in the literature, cannot force an alternative question interpretation or a yes/no question interpretation on their own. Several theories of disjunctive questions are discussed in the light of these experimental results.

동시인용정보를 이용한 동명이인 저자의 중의성 해소

동시인용은 서로 다른 두 연구가 이후의 새로운 연구에서 동시 인용되는 것이다. 이 연구는 동시인용과 저자식별의 관계를 다룬다. 저자식별은 문헌에 출현한 동명의 저자명들을 실 세계 저자로 식별하는 것이다. 동시인용은, 한 사람의 관련된 연구들이 이후 또 다른 연구들에서 타인 혹은 자신에 의해 동시 인용되는 증거를 수집함으로써, 저자식별의 절차와 성능에 영향을 미칠 수 있다. 이 연구는 구글 스칼라로부터 동시인용을 자동 수집하는 절차를 제시하고 동시인용 정보를 저자식별의 기존 자질들과 효율적으로 결합하는 새로운 군집알고리즘을 제안한다. 실험을 통해 동시인용이 저자식별에 미치는 긍정적 효과를 확인하였다. Co-citation means that two or more studies are cited together by a later study. This paper deals with the relationship between co-citation and author disambiguation. Author disambiguation is to cluster same-name author instances into real-world individuals. Co-citation may influence author disambiguation in terms that two or more related research works performed by the same person may be co-cited by some later studies. This article describes automated steps to gather co-citation information from Google scholar, and proposes a new clustering algorithm to effectively integrate co-citation information with other author disambiguation features. Experiments showed that co-citation helps to improve the performance of author disambiguation.

On co-authorship for author disambiguation

Author name disambiguation deals with clustering the same-name authors into different individuals. To attack the problem, many studies have employed a variety of disambiguation features such as coauthors, titles of papers/publications, topics of articles, emails/affiliations, etc. Among these, co-authorship is the most easily accessible and influential, since inter-person acquaintances represented by co-authorship could discriminate the identities of authors more clearly than other features. This study attempts to explore the net effects of co-authorship on author clustering in bibliographic data. First, to handle the shortage of explicit coauthors listed in known citations, a web-assisted technique of acquiring implicit coauthors of the target author to be disambiguated is proposed. Then, a coauthor disambiguation hypothesis that the identity of an author can be determined by his/her coauthors is examined and confirmed through a variety of author disambiguation experiments.

저자 식별을 위한 자질 비교

학술 정보에서 저자는, 실세계의 한 저자가 형태적으로 둘 이상의 저자명으로 출현할 수 있으며, 서로 다른 저자들이 동일한 저자명을 공유하기도 한다. 이는 각각 학술 정보에 대한 검색 및 탐색에 있어, 재현율과 정확률을 저하시키는 요인이다. 이 연구에서는 후자에 해당하는 저자의 동명이인 문제에 있어, 그 중의성 해소를 위한 자질의 특성에 집중하고자 한다. 최근까지, 저자 식별을 위한 자질로, 공저자, 논문 제목, 게재지명과 같은 서지 내적 자질과, 논문 원문 텍스트로부터 획득되는 전자메일주소, 소속기관, 논문의 토픽 등과 같은 서지 외적 자질이 사용되어 왔다. 그러나, 이러한 자질들이 저자 식별에 미치는 영향에 대한 비교 분석 연구는 찾아 보기 힘들다. 이 연구에서는, 한글 저자명에 대해 원문과 연계된 대용량 저자 식별 평가 셋을 구축하여, 동명 저자 중의성 해소에 있어 다양한 자질들의 특성을 비교한다. There exists a many-to-many mapping relationship between persons and their names. A person may have multiple names, and different persons may share the same name. These synonymous and homonymous names may severely deteriorate the recall and precision of the person search, respectively. This study addresses the characteristics of features for resolving homonymous author names appearing in citation data. As disambiguation features, previous works have employed citation-internal features such as co-authorship, titles of articles, titles of publications as well as citation-external features such as emails, affiliations, Web evidences. To the best of our knowledge, however, there has been no literature to deal with the influences of features on author disambiguation. This study analyzes the effect of individual features on author resolution using a large-scale test set for Korean.

Translation Disambiguation in Mixed Language Queries

Code-switching is very common among bilingual speakers. Spoken queries by these speakers are typically in mixed language. In this paper, we propose an unsupervised method for mixed-language query understanding, using only a monolingual corpus and a bilingual dictionary. Secondary-language words mixed in a primary-language query are translated into words in the primary language. We found that using a single disambiguation feature for translation is more effective than using multiple features, provided this feature is based on the most salient seed-word, chosen automatically by con?dence scoring. We propose and compare four types of disambiguation features that are based on context seed-words. A baseline method uses the nearest neighboring seed-word as disambiguation feature. Multiple-context seed-word voting is also proposed in order to enlarge the context window. On the other hand, merely using the inverse-distance as weights on context words degrades the performance as it runs counter to the potential underlying syntactic relations between words. Our ?nal proposal is a solution that uses multiple-context seed-words and the translation candidates of all mixed language words to select a single most salient seed-word for translation disambiguation. The translation disambiguation accuracy for this feature is at 83.7% for all words in the ATIS spontaneous speech query database, and 66.7% for content words.