Knowledge Models Used
Mark Doyle, *protected email*
Mark Doyle, *protected email*
PhySH (Physics Subject Headings) is a physics classification scheme developed by the American Physical Society to organize journal, meeting, and other content by topic, available for use starting in January 2016. It is intended initially to meet the specific goals of the APS, while a longer term goal is to make it available for use by the broader community. PhySH consists of hierarchies of concepts grouped into facets: Research Areas, Physical Systems, Properties, Techniques, and Professional Topics. The concepts are also organized by discipline for convenience. Individual concepts may belong to more than one facet or discipline.
We (Macmillan Science and Education) have made efforts to begin linking our domain models to external datasets. Our article-types, journals and subjects models are now linked to DBpedia (Wikipedia) and Wikidata. Our subjects model is additionally linked to Bio2RDF and MeSH.
Also, our core model is now linked to a number of other external models (CIDOC, FaBIO, schema.org, etc.).
Our models, now including these links, are available to view or download at nature.com/ontologies.
We will continue to refine and expand these links and would be interested in any thoughts, ideas and feedback from the community, particularly around any additional datasets we should consider linking to.
The NPG ArticleTypes Ontology is a categorization of kinds of publication which are used to index and group content published by Springer Nature. This taxonomy is organised into a single tree using the SKOS vocabulary. It includes article-types that are directly applied to content, such as Article, Review Article, News, or Book Review plus higher-level groupings such as Research, News and Comment, or Amendments and Corrections.
Applied manually by authors or editorial staff as part of the standard publishing workflow.
This model allows us to categorize content based on the type of publication, allowing content of similar type to be grouped or filtered at varying degrees of granularity.
We have a deep taxonomy of CS concepts – at the deepest level of the tree there are seven levels. The most useful concepts for precision search are of course the most granular concepts represented by the leaves of the tree.
However, concepts can be multi-parented, so the accurate application of a concept to a text requires that the context within the tree, i.e., the correct branch, be understood.
While expert authors who apply the terms to their articles have varying degrees of interest and attention to this indexing task, our experience shows that they rarely misapply terms – sometimes they appear lazy and are happy to assign only high-level concepts such as “Software” which is not too useful.
However, our experience with an auto-tagger shows that a huge amount of “noise” is created. We consider the noise unacceptable – presenting it to users will create distrust in the taxonomy itself.
We have been expanding the logical rules of the auto-tagger in an effort to reduce the noise to an acceptable level. So far, without success.
I have been trying to understand why.
So far, the best explanation I can come up with is that while hierarchical context is readily understood by the human brain, auto-taggers based on statistical occurrences of a concept and within proximity of other words and concepts, cannot accurately reproduce hierarchical context.
Any advice would be appreciated.
Springer Nature is one of the world’s leading global research, educational and professional publishers, home to an array of respected and trusted brands providing quality content through a range of innovative products and services.
Springer Nature is the world’s largest academic book publisher, publisher of the world’s most influential journals and a pioneer in the field of open research. Springer Nature was formed in 2015 through the merger of Nature Publishing Group, Palgrave Macmillan, Macmillan Education and Springer Science+Business Media.
Springer Nature is embracing linked data technologies as an integral part of its content publishing operations and has developed a data model which is highly responsive to new and legacy business requirements. Linked data is central to the customer experience in providing content discovery applications and in facilitating emergent behaviours in interacting with content.
Many of the models developed have recently been published on our Ontologies Portal at nature.com/ontologies and are shared in order to contribute to the wider linked data community and to provide a public reference. These models cover publication things – articles, figures, etc. – and classification things – article-types, subjects, etc. – plus additional things used to manage our content publishing operation – assets, events, etc.
We have also published a model for conference proceedings on our LOD Conference Portal at lod.springer.com
The NPG Subjects Ontology is a polyhierarchical categorization of scholarly subject areas which are used for the indexing of content by Springer Nature. It includes subject terms of varying levels of specificity such as Biological sciences (top level), Cancer (level 2), or B-2 cells (level 7). In total there are more than 2750 subject terms, organised into a polyhierarchical tree using the SKOS vocabulary.
Applied manually, by authors or editorial staff as part of the standard publishing workflow, or by professional indexers
The NPG Subjects Ontology constitutes the main backbone of nature.com subject areas, a new section on nature.com that allows users to browse content topically rather than navigate via the more usual journal paradigm. Each of the terms in the ontology includes a link to the relevant subject page on nature.com.
Thesaurus of the National Information Center for Educational Media (NICEM), used for indexing the records of NICEM’s bibliographic database.
Used for indexing bibliographic records of non-print educational media in the NICEM database.
The thesaurus contains terms that reflect the subject matter of educational material, especially at the K-12 levels. An associated rule base that has been developed specifically for those terms enables appropriate indexing and accurate retrieval of bibliographic records, as well as user-friendly browsing in conjunction with a search interface.
PLOS (Public Library of Science) is a nonprofit publisher and advocacy organization founded to accelerate progress in science and medicine by leading a transformation in research communication.
Our core objectives are to provide ways to overcome unnecessary barriers to immediate availability, access and use of research, pursue a publishing strategy that optimizes the quality and integrity of the publication process, and develop innovative approaches to the assessment, organization and reuse of ideas and data.
An indexing system for the newspaper industry. A specialized group of terms with the newspaper industry’s indexing needs in mind. The vocabulary is divided into sections that correspond to the sections of a typical newspaper. An accompanying rule base enables highly accurate categorization of newspaper articles.
Customized version used by Acquire Media for categorization of news items, and RSS delivery according to customers’ interests.
Categorization of news stories (including archived stories) by and for newspaper publishers; indexing of 20th and 21st century historical studies.
Every news day, you can tag the articles as they are produced through a cloud service or installed on your own local servers. We automatically feed this data through NewsIndexer, which scans every article and searches for terms similar to those in its controlled vocabulary. NewsIndexer then displays these terms for the human indexer’s review and approval. For backfile collections you can just accept the indexing as an automatic batch process. For ongoing daily feeds you might want to review all or a random sample of the results on a regular basis for maintenance.