Search

We propose a method for online news stream clustering that is a variant of the non-parametric streaming K-means algorithm. Our model uses a combination of sparse and dense document representations, aggregates document-cluster similarity along these multiple representations and makes the clustering decision using a neural classifier. The weighted document-cluster similarity model is learned using a novel adaptation of the triplet loss into a linear classification objective. We show that the use of a suitable fine-tuning objective and external knowledge in pre-trained transformer models yields significant improvements in the effectiveness of contextual embeddings for clustering. Our model achieves a new state-of-the-art on a standard stream clustering dataset of English documents., Comment: To appear in Proceedings of The 16th Conference of the European Chapter of the Association for Computational Linguistics

The large scale of scholarly publications poses a challenge for scholars in information seeking and sensemaking. Bibliometric, information retrieval (IR), text mining, and natural language processing techniques can assist to address this challenge, but have yet to be widely used in digital libraries (DL). This special issue on bibliometric-enhanced information retrieval and natural language processing for digital libraries (BIRNDL) was compiled after the first joint BIRNDL workshop that was held at the joint conference on digital libraries (JCDL 2016) in Newark, New Jersey, USA. It brought together IR and DL researchers and professionals to elaborate on new approaches in natural language processing, information retrieval, scientometric, and recommendation techniques that can advance the state of the art in scholarly document understanding, analysis, and retrieval at scale. This special issue includes 14 papers: four extended papers originating from the first BIRNDL workshop 2016 and the BIR workshop at ECIR 2016, four extended system reports of the CL-SciSumm Shared Task 2016 and six original research papers submitted via the open call for papers.

We describe the participation and the official results of the 2nd Computational Linguistics Scientific Summarization Shared Task (CL-SciSumm), held as a part of the BIRNDL workshop at the Joint Conference for Digital Libraries 2016 in Newark, New Jersey. CL-SciSumm is the first medium-scale Shared Task on scientific document summarization in the computational linguistics (CL) domain. Participants were provided a training corpus of 30 topics, each comprising of a reference paper (RP) and 10 or more citing papers, all of which cite the RP. For each citation, the text spans (i.e., citances) that pertain to the RP have been identified. Participants solved three sub-tasks in automatic research paper summarization using this text corpus. Fifteen teams from six countries registered for the Shared Task, of which ten teams ultimately submitted and presented their results. The annotated corpus comprised 30 target papers--currently the largest available corpora of its kind. The corpus is available for free download and use at https://github.com/WING-NUS/scisumm-corpus.

Big Science and cross-disciplinary collaborations have reshaped the intellectual structure of research areas. A number of works have tried to uncover this hidden intellectual structure by analyzing citation contexts. However, none of them analyzed by document logical structures such as sections. The two major goals of this study are to find characteristics of authors who are highly cited section-wise and to identify the differences in section-wise author networks. This study uses 29,158 of research articles culled from the ACL Anthology, which hosts articles on computational linguistics and natural language processing. We find that the distribution of citations across sections is skewed and that a different set of highly cited authors share distinct academic characteristics, according to their citation locations. Furthermore, the author networks based on citation context similarity reveal that the intellectual structure of a domain differs across different sections.

The large scale of scholarly publications poses a challenge for scholars in information seeking and sense-making. Bibliometrics, information retrieval (IR), text mining, and NLP techniques could help in these activities, but are not yet widely implemented in digital libraries. The 2nd joint BIRNDL workshop was held at the 40th ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2017) in Tokyo, Japan. BIRNDL 2017 intended to stimulate IR researchers and digital library professionals to elaborate on new approaches in natural language processing, information retrieval, scientometric, and recommendation techniques that can advance the state-of-the-art in scholarly document understanding, analysis, and retrieval at scale. The workshop incorporated three paper sessions and the 3rd edition of the CL-SciSumm Shared Task.

Next to keeping up with the growing literature in their own and related fields, scholars increasingly also need to rebut pseudo-science and disinformation. To address these challenges, computational work on enhancing search, summarization, and analysis of scholarly documents has flourished. However, the various strands of research on scholarly document processing remain fragmented. To reach to the broader NLP and AI/ML community, pool distributed efforts and enable shared access to published research, we held the 1st Workshop on Scholarly Document Processing at EMNLP 2020 as a virtual event. The SDP workshop consisted of a research track (including a poster session), two invited talks and three Shared Tasks (CL-SciSumm, Lay-Summ and LongSumm), geared towards easier access to scientific methods and results. Website: https://ornlcda.github.io/SDProc

We present the results of three Shared Tasks held at the Scholarly Document Processing Workshop at EMNLP2020: CL-SciSumm, LaySumm and LongSumm. We report on each of the tasks, which received 18 submissions in total, with some submissions addressing two or three of the tasks. In summary, the quality and quantity of the submissions show that there is ample interest in scholarly document summarization, and the state of the art in this domain is at a midway point between being an impossible task and one that is fully resolved.

The large scale of scholarly publications poses a challenge for scholars in information seeking and sensemaking. Information retrieval~(IR), bibliometric and natural language processing (NLP) techniques could enhance scholarly search, retrieval and user experience but are not yet widely used. To this purpose, we propose the third iteration of the Joint Workshop on Bibliometric-enhanced Information Retrieval and Natural Language Processing for Digital Libraries (BIRNDL). The workshop is intended to stimulate IR, NLP researchers and Digital Library professionals to elaborate on new approaches in natural language processing, information retrieval, scientometrics, text mining and recommendation techniques that can advance the state-of-the-art in scholarly document understanding, analysis, and retrieval at scale. The BIRNDL workshop will incorporate multiple invited talks, paper sessions, a poster session and the 4th edition of the Computational Linguistics (CL) Scientific Summarization Shared Task.

We systematically confirm that instructors are strongly influenced by the user interface presentation of Massive Online Open Course (MOOC) discussion forums. In a large scale dataset, we conclusively show that instructor interventions exhibit strong position bias, as measured by the position where the thread appeared on the user interface at the time of intervention. We measure and remove this bias, enabling unbiased statistical modelling and evaluation. We show that our de-biased classifier improves predicting interventions over the state-of-the-art on courses with sufficient number of interventions by 8.2% in F1 and 24.4% in recall on average.

The large scale of scholarly publications poses a challenge for scholars in information seeking and sensemaking. Bibliometrics, information retrieval (IR), text mining and NLP techniques could help in these search and look-up activities, but are not yet widely used. This workshop is intended to stimulate IR researchers and digital library professionals to elaborate on new approaches in natural language processing, information retrieval, scientometrics, text mining and recommendation techniques that can advance the state-of-the-art in scholarly document understanding, analysis, and retrieval at scale. The BIRNDL workshop at SIGIR 2017 will incorporate an invited talk, paper sessions and the third edition of the Computational Linguistics (CL) Scientific Summarization Shared Task., Comment: 2 pages, workshop paper accepted at the SIGIR 2017

International audience; The large scale of scholarly publications poses a challenge for scholars in information-seeking and sensemaking. Bibliometric, information retrieval (IR), text mining and NLP techniques could help in these activities, but are not yet widely used in digital libraries. This workshop is intended to stimulate IR researchers and digital library professionals to elaborate on new approaches in natural language processing, information retrieval, scientometric and recommendation techniques which can advance the state-of-the-art in scholarly document understanding, analysis and retrieval at scale.

We address the tasks of recovering bibliographic and document structure metadata from scholarly documents. We leverage higher order semi-Markov conditional random fields to model long-distance label sequences, improving upon the performance of the linear-chain conditional random field model. We introduce the notion of extensible features, which allows the expensive inference process to be simplified through memoization, resulting in lower computational complexity. Our method significantly betters the state-of-the-art on three related scholarly document extraction tasks.

We propose WordNews, a web browser extension that allows readers to learn a second language vocabulary while reading news online. Injected tooltips allow readers to look up selected vocabulary and take simple interactive tests. We discover that two key system components needed improvement, both which stem from the need to model context. These two issues are real-world word sense disambiguation (WSD) to aid translation quality and constructing interactive tests. For the first, we start with Microsoft’s Bing translation API but employ additional dictionary-based heuristics that significantly improve translation in both coverage and accuracy. For the second, we propose techniques for generating appropriate distractors for multiple-choice word mastery tests. Our preliminary user survey confirms the need and viability of such a language learning platform.

We introduce Enlil, an information extraction system that discovers the institutional affiliations of authors in scholarly papers. Enlil consists of two steps: one that first identifies authors and affiliations using a conditional random field; and a second support vector machine that connects authors to their affiliations. We benchmark Enlil in three separate experiments drawn from three different sources: the ACL Anthology Corpus, the ACM Digital Library, and a set of cross-disciplinary scientific journal articles acquired by querying Google Scholar. Against a state-of-the-art production baseline, Enlil reports a statistically significant improvement in F_1 of nearly 10% (p 90%) and automatically-acquired input (F_1 > 80%).We have deployed Enlil in a case study involving Asian genomics research publication patterns to understand how government sponsored collaborative links evolve. Enlil has enabled our team to construct and validate new metrics to quantify the facilitation of research as opposed to direct publication.