141 research outputs found
Semantic annotation of multilingual learning objects based on a domain ontology
Get PDFOne of the important tasks in the use of learning resources in e-learning is the necessity to annotate learning objects with appropriate metadata. However, annotating resources by hand is time consuming and difficult. Here we explore the problem of automatic extraction of metadata for description of learning resources. First, theoretical constraints for gathering certain types of metadata important for e-learning systems are discussed. Our approach to annotation is then outlined. This is based on a domain ontology, which allows us to annotate learning resources in a language independent way.We are motivated by the fact that the leading providers of learning content in various domains are often spread across countries speaking different languages. As a result, cross-language annotation can facilitate accessibility, sharing and reuse of learning resources
Recommended from our members
Incidental or influential? – A decade of using text-mining for citation function classification.
Get PDFThis work looks in depth at several studies that have attempted to automate the process of citation importance classification based on the publications’ full text. We offer a comparison of their individual similarities, strengths and weaknesses. We analyse a range of features that have been previously used in this task. Our experimental results confirm that the number of in-text references are highly predictive of influence. Contrary to the work of Valenzuela et al. (2015), we find abstract similarity one of the most predictive features. Overall, we show that many of the features previously described in literature have been either reported as not particularly predictive, cannot be reproduced based on their existing descriptions or should not be used due to their reliance on external changing evidence. Additionally, we find significant variance in the results provided by the PDF extraction tools used in the pre-processing stages of citation extraction. This has a direct and significant impact on the classification features that rely on this extraction process. Consequently, we discuss challenges and potential improvements in the classification pipeline, provide a critical review of the performance of individual features and address the importance of constructing a large-scale gold-standard reference dataset
Recommended from our members
Can we do better than co-citations? Bringing Citation Proximity Analysis from idea to practice in research articles recommendation
Get PDFIn this paper, we build on the idea of Citation Proximity Analysis (CPA), originally introduced in [1], by developing a step by step scalable approach for building CPA-based recommender systems. As part of this approach, we introduce three new proximity functions, extending the basic assumption of co-citation analysis (stating that the more often two articles are co-cited in a document, the more likely they are related) to take the distance between the co-cited documents into account. Ask- ing the question of whether CPA can outperform co-citation analysis in recommender systems, we have built a CPA based recommender system from a corpus of 368,385 full-texts articles and conducted a user survey to perform an initial evaluation. Two of our three proximity functions used within CPA outperform co-citations on our evaluation dataset
From open access metadata to open access content: two principles for increased visibility of open access content
Get PDFAn essential goal of the open access (OA) movement is free availability of research outputs on the Internet. One of the recommended ways to achieve this is through open access repositories (BOAI, 2002). Given the growing number of repositories and the significant proportion of research outputs already available as OA (Laakso & Bjork, 2012), it might come as a surprise that OA content is not necessarily easily discoverable on the Internet (Morrisson, 2012; Konkiel, 2012), more precisely, it is available, but often difficult to find. If OA content in repositories cannot be discovered, there is little incentive to make it available on the Internet in the first place. Therefore, not trying hard enough to increase the visibility of OA content would be a lost opportunity for achieving the main OA goals, including also the reuse potential of OA content. In this paper, we build on our experience in finding and aggregating open access content (not just metadata) from repositories, discussing the main issues and summarizing the lessons learned into two principles that, if adopted, will dramatically increase the discoverability of OA content on the Internet and will improve the possibilities of OA content reuse
Extraction of semantic relations from texts
Get PDFIn recent years the amount of unstructured data stored on the Internet and other digital sources has increased significantly. These data contain often valuable, but hardly retrievable information. The term unstructured data refers mainly to data that does not have a data structure. As a result of this, the unstructured data is not easily readable by machines. In this work, we present a simple method for automatic extraction of semantic relations that can be used to precisely locate valuable pieces of information
Information Extraction from Biomedical Texts
Get PDFV poslední době bylo vynaloženo velké úsilí k tomu, aby byly biomedicínské znalosti, typicky uložené v podobě vědeckých článků, snadněji přístupné a bylo možné je efektivně sdílet. Ve skutečnosti ale nestrukturovaná podstata těchto textů způsobuje velké obtíže při použití technik pro získávání a vyvozování znalostí. Anotování entit nesoucích jistou sémantickou informaci v textu je prvním krokem k vytvoření znalosti analyzovatelné počítačem. V této práci nejdříve studujeme metody pro automatickou extrakci informací z textů přirozeného jazyka. Dále zhodnotíme hlavní výhody a nevýhody současných systémů pro extrakci informací a na základě těchto znalostí se rozhodneme přijmout přístup strojového učení pro automatické získávání exktrakčních vzorů při našich experimentech. Bohužel, techniky strojového učení často vyžadují obrovské množství trénovacích dat, která může být velmi pracné získat. Abychom dokázali čelit tomuto nepříjemnému problému, prozkoumáme koncept tzv. bootstrapping techniky. Nakonec ukážeme, že během našich experimentů metody strojového učení pracovaly dostatečně dobře a dokonce podstatně lépe než základní metody. Navíc v úloze využívající techniky bootstrapping se podařilo významně snížit množství dat potřebných pro trénování extrakčního systému.Recently, there has been much effort in making biomedical knowledge, typically stored in scientific articles, more accessible and interoperable. As a matter of fact, the unstructured nature of such texts makes it difficult to apply knowledge discovery and inference techniques. Annotating information units with semantic information in these texts is the first step to make the knowledge machine-analyzable. In this work, we first study methods for automatic information extraction from natural language text. Then we discuss the main benefits and disadvantages of the state-of-art information extraction systems and, as a result of this, we adopt a machine learning approach to automatically learn extraction patterns in our experiments. Unfortunately, machine learning techniques often require a huge amount of training data, which can be sometimes laborious to gather. In order to face up to this tedious problem, we investigate the concept of weakly supervised or bootstrapping techniques. Finally, we show in our experiments that our machine learning methods performed reasonably well and significantly better than the baseline. Moreover, in the weakly supervised learning task we were able to substantially bring down the amount of labeled data needed for training of the extraction system.
Recommended from our members
Analyzing Citation-Distance Networks for Evaluating Publication Impact
Get PDFStudying citation patterns of scholarly articles has been of interest to many researchers from various disciplines. While the relationship of citations and scientific impact has been widely studied in the literature, in this paper we develop the idea of analyzing the semantic distance of scholarly articles in a citation network (citation-distance network) to uncover patterns that reflect scientific impact. More specifically, we compare two types of publications in terms of their citation-distance patterns, seminal publications and literature reviews, and focus on their referencing patterns as well as on publications which cite them. We show that seminal publications are associated with a larger semantic distance, measured using the content of the articles, between their references and the citing publications, while literature reviews tend to cite publications from a wider range of topics. Our motivation is to understand and utilize this information to create new research evaluation metrics which would better reflect scientific impact
My repository is being aggregated: a blessing or a curse?
Get PDFUsage statistics are frequently used by repositories to justify their value to the management who decide about the funding to support the repository infrastructure. Another reason for collecting usage statistics at repositories is the increased use of webometrics in the process of assessing the impact of publications and researchers. Consequently, one of the worries repositories sometimes have about their content being aggregated is that they feel aggregations have a detrimental effect on the accuracy of statistics they collect. They believe that this potential decrease in reported usage can negatively influence the funding provided by their own institutions. This raises the fundamental question of whether repositories should allow aggregators to harvest their metadata and content. In this paper, we discuss the benefits of allowing content aggregations harvest repository content and investigate how to overcome the drawbacks
Recommended from our members
Linking Textual Resources to Support Information Discovery
Get PDFA vast amount of information is today stored in the form of textual documents, many of which are available online. These documents come from different sources and are of different types. They include newspaper articles, books, corporate reports, encyclopedia entries and research papers. At a semantic level, these documents contain knowledge, which was created by explicitly connecting information and expressing it in the form of a natural language. However, a significant amount of knowledge is not explicitly stated in a single document, yet can be derived or discovered by researching, i.e. accessing, comparing, contrasting and analysing, information from multiple documents. Carrying out this work using traditional search interfaces is tedious due to information overload and the difficulty of formulating queries that would help us to discover information we are not aware of.
In order to support this exploratory process, we need to be able to effectively navigate between related pieces of information across documents. While information can be connected using manually curated cross-document links, this approach not only does not scale, but cannot systematically assist us in the discovery of sometimes non-obvious (hidden) relationships. Consequently, there is a need for automatic approaches to link discovery.
This work studies how people link content, investigates the properties of different link types, presents new methods for automatic link discovery and designs a system in which link discovery is applied on a collection of millions of documents to improve access to public knowledge
What Others Say About This Work? Scalable Extraction of Citation Contexts from Research Papers
Get PDFThis work presents a new, scalable solution to the problem of extracting citation contexts: the textual fragments surrounding citation references. These citation contexts can be used to navigate digital libraries of research papers to help users in deciding what to read. We have developed a prototype system which can retrieve, on-demand, citation contexts from the full text of over 15 million research articles in the Mendeley catalog for a given reference research paper. The evaluation results show that our citation extraction system provides additional functionality over existing tools, has two orders of magnitude faster runtime performance, while providing a 9% improvement in F-measure over the current state-of-the-art
- …
