Dynamic author name disambiguation for growing digital libraries

This paper tackles the well known problem of Author Name Disambiguation (AND) in Digital Libraries (DL). Following [14,13], we assume that an individual tends to create a distinctively coherent body of work that can hence form a single cluster containing all of his/her articles yet distinguishing them from those of everyone else with the same name. Still, we believe the information contained in a DL may be not sufficient to allow an automatic detection of such clusters; this lack of information becomes even more evident in federated digital libraries, where the labels assigned by librarians may belong to different controlled vocabularies or different classification systems, and in digital libraries on the web where records may be not assigned neither subject headings nor classification numbers. Hence, we exploit Topic Models, extracted from Wikipedia, to enhance records metadata and use Agglomerative Clustering to disambiguate ambiguous author names by clustering together similar records; records in different clusters are supposed to have been written by different people. We investigate the following two research questions: (a) are the Classification Systems and Subject Heading labels manually assigned by librarians general and informative enough to disambiguate Author Names via clustering techniques? (b) Do Topic Models induce from large corpora the conceptual information necessary for labelling automatically DL metadata and grasp topic similarities of the records? To answer these questions, we will use the Library Catalogue of the Bolzano University Library as case study.KeywordsDigital LibrariesTopic ModelsAuthor Name Disambiguation

Digital bibliographic repositories, including publications, authors, and research fields are essential for sharing scientific information. Nevertheless, the information retrieval, extraction, and classification efficiency in such archives is threatened by author name ambiguity. This paper addresses the Author Name Disambiguation (AND) problem by proposing a hybrid machine learning method integrating Bidirectional Encoder Representations from Transformers (BERT), Graph Convolutional Network (GCN), and Graph Enhanced Hierarchical Agglomerative Clustering (GHAC) approaches. The BERT model extracts textual data from scientific documents, the GCN structures global data from academic graphs, and GHAC considers heterogeneous networks’ global context to identify scientific collaboration patterns. We compare the hybrid method with AND state-of-the-art work using a publicly accessible data set consisting of 7,886 documents, 137 unique authors, and 14 groups of ambiguous authors, along with recognized validation metrics. The results achieved a high precision score of 93.8%, recall of 96.3%, F1-measure of 95%, Average Cluster Purity (ACP) of 96.5%, Average Author Purity (AAP) of 97.4% and K-Metric of 96.9%. Compared to the AND baseline approach, the hybrid method presents better results indicating a promising approach.

The problem with Author Name Disambiguation is to determine whether the same name in the bibliographic archive refers to the same author or not. Currently, author identification on The Labeled Digital Bibliography and Library Project (DBLP) is triggered by a request for an author who finds his publication mixed with other people's writing. Name ambiguity leads to incorrect identification and attribution of credit to authors. Despite much research in the last decade, the issue of ambiguity of the author's name remains largely unsolved. In this paper, the Capsule Networks (CapsNets) method is proposed to resolve the ambiguity of the author's name. The proposed method obtains the best accuracy in four Name Disambiguation problems including homonyms, synonyms, and non-homonyms synonyms, which is an average of 99% on training and testing data. Likewise, the overall data tested has an accuracy of 99.83% with a low error value. In addition, CapsNets were tested with Performance Measurements including Sensitivity, Precision, and F1-Score. Capsnets can identify authors in DBLP bibliographic data by using a number of attributes such as author name, co-author, venue, title, and year.

Information for Contributors

Author name ambiguity has been a long-standing problem which impairs the accuracy of publication retrieval and bibliometric methods. Most of the existing disambiguation methods are built on similarity measures, e.g., “Jaccard Coefficient”, between two sets of papers to be disambiguated, each set represented by a set of categorical features, e.g., coauthors and published venues. Such measures perform bad when the two sets are small, which is typical in Author Name Disambiguation. In this paper, we propose a novel categorical set similarity measure. We model an author’s preference, e.g., to venues, using a categorical distribution, and derive a likelihood ratio to estimate the likelihood that the two sets are drawn from the same distribution. This likelihood ratio is used as the similarity measure to decide whether two sets belong to the same author. This measure is mathematically principled and verified to perform well even when the cardinalities of the two compared sets are small. Additionally, we propose a new method to estimate the number of distinct authors for a given name based on the name statistics extracted from a digital library. Experiment shows that our method significantly outperforms a baseline method, a widely used benchmark method, and a real system.KeywordsName DisambiguationCategorical Sampling Likelihood Ratio

Authors' names are a critical bibliographic element when searching or browsing academic articles stored in digital libraries. However, extracting such bibliographic data from printed documents requires human intervention; it is therefore not cost-effective, even using various document image-processing techniques such as Optical Character Recognition (OCR). In this paper, we describe an automatic authors' names extraction method for academic articles scanned with OCR mark-up. The proposed method first extracts authors' blocks, which include assumed author/delimiter characters based on layout analysis, and then uses a specifically designed Hidden Markov Model (HMM) for labeling the unsegmented character strings in the block as those of either an author or a delimiter. We applied the proposed method to Japanese academic articles. Results of these experiments showed that the proposed method correctly extracted more than 99% of authors' blocks with manual tuning; the proposed HMM correctly labeled more than 95% of the author name strings.

Name ambiguity is a common problem in many bibliographic repositories affecting data integrity and validity. This article presents an author name disambiguation (AND) literature review using the theory of the consolidated meta-analytic approach, including quantitative techniques and bibliometric aspects. The literature review covers information from 211 documents of the Web of Science and Scopus databases in the period 2003 to 2022. A taxonomy based on the literature was used to organize the identified approaches to solve the AND problem. We identified that the most widely used AND solving approaches are author grouping associated with similarity functions and clustering methods and some works using author assignment allied to classification methods. The countries that publish most in AND are the USA, China, Germany, and Brazil with 21%, 19%, 13% and 8% of the total papers, respectively. The review results provide an overview of AND state-of-the-art research that can direct further investigation based on the quantitative and qualitative information from the AND research history.

Entity disambiguation is an important step in many information retrieval applications. This paper proposes new research for entity disambiguation with the focus of name disambiguation in digital libraries. In particular, pairwise similarity is first learned for publications that share the same author name string (ANS) and then a novel Hierarchical Agglomerative Clustering approach with Adaptive Stopping Criterion (HACASC) is proposed to adaptively cluster a set of publications that share a same ANS to individual clusters of publications with different author identities. The HACASC approach utilizes a mixture of kernel ridge regressions to intelligently determine the threshold in clustering. This obtains more appropriate clustering granularity than non-adaptive stopping criterion. We conduct a large scale empirical study with a dataset of more than 2 million publication record pairs to demonstrate the advantage of the proposed HACASC approach.

In several author name disambiguation studies, some ethnic name groups such as East Asian names are reported to be more difficult to disambiguate than others. This implies that disambiguation approaches might be improved if ethnic name groups are distinguished before disambiguation. We explore the potential of ethnic name partitioning by comparing performance of four machine learning algorithms trained and tested on the entire data or specifically on individual name groups. Results show that ethnicity‐based name partitioning can substantially improve disambiguation performance because the individual models are better suited for their respective name group. The improvements occur across all ethnic name groups with different magnitudes. Performance gains in predicting matched name pairs outweigh losses in predicting nonmatched pairs. Feature (e.g., coauthor name) similarities of name pairs vary across ethnic name groups. Such differences may enable the development of ethnicity‐specific feature weights to improve prediction for specific ethic name categories. These findings are observed for three labeled data with a natural distribution of problem sizes as well as one in which all ethnic name groups are controlled for the same sizes of ambiguous names. This study is expected to motive scholars to group author names based on ethnicity prior to disambiguation.

How can we evaluate the performance of a disambiguation method implemented on big bibliographic data? This study suggests that the open researcher profile system, ORCID, can be used as an authority source to label name instances at scale. This study demonstrates the potential by evaluating the disambiguation performances of Author-ity2009 (which algorithmically disambiguates author names in MEDLINE) using 3 million name instances that are automatically labeled through linkage to 5 million ORCID researcher profiles. Results show that although ORCID-linked labeled data do not effectively represent the population of name instances in Author-ity2009, they do effectively capture the ‘high precision over high recall’ performances of Author-ity2009. In addition, ORCID-linked labeled data can provide nuanced details about the Author-ity2009’s performance when name instances are evaluated within and across ethnicity categories. As ORCID continues to be expanded to include more researchers, labeled data via ORCID-linkage can be improved in representing the population of a whole disambiguated data and updated on a regular basis. This can benefit author name disambiguation researchers and practitioners who need large-scale labeled data but lack resources for manual labeling or access to other authority sources for linkage-based labeling. The ORCID-linked labeled data for Author-ity2009 are publicly available for validation and reuse.

In the process of author name disambiguation (AND), varying characteristics and noise of different blocks significantly impact disambiguation performance. In this paper, we propose a block‐based adaptive hyperparameter optimization method that assigns optimal hyperparameters to each block without altering the original AND model structure. Based on this, a random forest model is trained using the optimized results to fit the relationship between the block's data features and its optimal hyperparameters, thereby enabling the prediction of hyperparameters for new blocks. Empirical studies on 6 state‐of‐the‐art AND algorithms, 11 public datasets, and a manually labeled dataset of China's information and communication technology (ICT) industry patents demonstrate that the proposed method significantly outperforms the original algorithms across multiple standard performance evaluation metrics (Cluster F1/Pairwise F1, B‐Cubed F1, and K metrics). The results of the random forest regression indicate that the selected 16 features effectively predict the optimal hyperparameters. Further analysis reveals a power‐law relationship between relative block size and both relative performance and relative optimized performance across all datasets and evaluation metrics, and the relative performance improvement of the adaptive hyperparameter optimization algorithm is particularly significant for smaller blocks. These findings provide theoretical support and practical guidance for the development of AND algorithms.

This paper addresses the author disambiguation problem in academic social network, namely, resolves the phenomenon of synonym problem “multiple names refer to one person” and polysemy problem “one name refers to multiple persons”. A unified semi-supervised framework is proposed to deal with both the synonym and polysemy problems. First, the framework uses semi-supervised approach to solve the cold-start problem in author disambiguation. Second, robust training data generating method based on multi-aspect similarity indicator is used and a way based on support vector machine is employed to model different kinds of feature combinations. Third, a self-taught procedure is proposed to solve ambiguity in coauthor information to boost the performances from other models. The proposed framework is verified on a large-scale real-world dataset, and obtains promising results.

Name ambiguity in the context of bibliographic citation records is a hard problem that affects the quality of services and content in digital libraries and similar systems. This problem occurs when an author publishes works under distinct names or distinct authors publish works under similar names. The challenges of dealing with author name ambiguity have led to a myriad of name disambiguation methods. In this tutorial, we characterize such methods by means of a proposed taxonomy, present an overview of some of the most representative ones and discuss open challenges.

서지 데이터베이스를 이용한 인용분석연구를 진행하기 이전에 이루어져야 할 과정 중 하나가 모호한 저자명의 식별이라고 할 수 있다. 대부분 서지 데이터베이스에는 저자의 성(姓)과 이름의 이니셜만을 표기하는 경우가 많은데, 중국이나 한국 등 아시아 국가 출신의 연구자들은 같은 성을 가진 사람이 매우 많고, 이름의 이니셜까지 같은 경우도 상당히 많아서 이름검색만으로 찾고자 하는 저자를 식별해내기가 쉽지 않기 때문이다. 아시아 국가 출신의 학자들이 유난히 많은 연구분야들에서는 이러한 문제들이 더더욱 큰 문제가 되며, 인용분석 뿐만 아니라 일반적인 정보검색에서도 매우 중요한 요인이 될 수 있다. 모호한 저자명을 식별해내는 방법에는 자동화된 알고리듬을 이용하여 각각의 저자를 식별해내는 방법과 저자 클러스터링을 얻어내기 위해 일일이 수작업으로 데이터셋을 구축하는 방법, 그리고 두 가지 방법을 혼용한 반자동화된 방법 등이 있다. 본 연구는 "모호한 저자명 식별"을 위해 개발된 여러 가지 방법들을 고찰해보기로 한다. In citation analysis, author names are often used as the unit of analysis and some authors are indexed under the same name in bibliographic databases where the citation counts are obtained from. There are many techniques for author name disambiguation, using supervised, unsupervised, or semisupervised learning algorithms. Unsupervised approach uses machine learning algorithms to extract necessary bibliographic information from large-scale databases and digital libraries, while supervised approaches use manually built training datasets for clustering author groups for combining them with learning algorithms for author name disambiguation. The study examines various techniques for author name disambiguation in the hope for finding an aid to improve the precision of citation counts in citation analysis, as well as for better results in information retrieval.

In digital libraries, ambiguous author names may occur because of the existence of multiple authors with the same name or different name variations for the same person. In recent years, name disambiguation has become a major challenge when integrating data from multiple sources in bibliographic digital libraries. Most of the previous works solve this issue by using many attributes, such as coauthors, title of articles/publications, topics of articles, and years of publications. However, in most cases, we can only get the coauthor and title attributes. In this paper, we propose an approach which is based on Hierarchical Agglomerative Clustering (HAC) and only use the coauthor and title attributes, but can more effectively identify the disambiguation authors. The whole algorithm can divide into two stages. In the first stage, we employ a pair-wise grouping algorithm which is based on coauthors’name to group records into clusters. Then, we merge two clusters if the similarity of the article titles from two clusters reach the threshold. Here, we use three kinds of similarity algorithms such as Jaccard Similarity, Cosine Similarity and Euclidean Distance to compare the similarity between the titles of two clusters. To minimize the risk of using only one similarity metric, we design the concept of ranking confidence to measure the confidence of different similarity meausrements. The ranking confidence decides which similarity measure to use when merging clusters. In the experiments, we use PairPresicion, PairRecall and PairF1 score to evaluate our method and compare with other methods. Experimental results indicate that our method significantly outperforms the baseline methods: HAC, K-means and SACluster when only use coauthor and title attributes.