ScienceWISE phase 2 continuation

Long Title: Web-based Interactive Semantic Environment for e-Science - Phase 2 continuation
EPF Lausanne
Domain: ELS
Status: finished
Start Date: 01.01.2013
End Date: 30.04.2013
Project Leader: A. Boyarsky
Website: http://sciencewise.info

ScienceWISE - platform for creation of virtual organizations of scientists, working together on a dynamical generation of professional field-specific ontologies - was further extended with focus on semantic bookmarking and scientific ontology.

(See also Phase 1 and Phase 2)


The following resources were developed:

Component Description

ScienceWISE portal
SPARQL endpoint The SPARQL endpoint enables users (human or other) to query the ScienceWISE knowledge base via the SPARQL language.
RDF/N-Triples access to ScienceWISE data Categories and metadata for N-Triples access. (N-Triples is a format for storing and transmitting data. It's a line-based, plain text serialisation format for RDF (Resource Description Framework) graphs.)
Tag Recommendation for Large-Scale Ontology-Based Information Systems Article in International SemanticWeb Conference (ISWC) 2012. Authors: R. Prokofyev, A. Boyarsky, O., Ruchayskiy, K. Aberer, G. Demartini, P. Cudre-Mauroux
Ontology-Based Word Sense Disambiguation in the Scientific Domain Accepted full papers for short oral and poster presentation at the 35th European Conference on Information Retrieval (ECIR) 2013. Authors: R. Prokofyev, G. Demartini, P. Cudre-Mauroux, A. Boyarsky, O. Ruchayskiy
From scientific papers to the scientific ontology: dynamical clustering of heterogeneous graphs and ontology crowdsourcing Workshop paper in Joint Workshop on Large and Heterogeneous Data and Quantitative Formalization in the Semantic Web (LHD+SemQuant 2012). Authors: A. Boyarsky, O. Ruchayskiy, Z. Yang, O. Zozulya, M. Charlaganov, P. De Los Rios
ScienceWISE: A Web-based Interactive Semantic Platform for Paper Annotation and Ontology Editing Demo paper in Extended Semantic Web Conference (ESWC) 2012. Authors: A. Astafiev, R. Prokofyev, C. Guéret, A. Boyarsky, O. Ruchayskiy

During phase 2 and the extension several planned new functions and data have been integrated.
ScienceWISE saves the researchers' time by automating some of their daily activities. It does so through semantic recommendation system, semantic bookmarking, and annotations. When annotating the system identifies the relevant concepts and produces a PDF of the manuscript with hyperlinks to relevant definitions/resources, thus providing additional details, comments or pedagogical materials.
Users can bookmark any paper from the major scientific pre-print repository (arXiv.org, CERN Document Server or NASA bibliographic system). The system automatically selects the most relevant concepts for bookmarking, to be further fine-tuned by the user. It allows to classify bookmarked papers, create collections and easily navigate to any bookmarked paper.
Scientists and young researchers therefore can cope with the large volume of daily publications in pre-print repositories or in open-access journals.
Using the system, each scientist adds new concepts, definitions, resources to the ScienceWISE ontology. Their contributions become available for everyone else. ScienceWISE is therefore a system that enables implicit crowd-sourcing of the expert knowledge.
The system is also a source of unique "experimental" data for research in key-phrase-extraction, topic modeling, and in complexity sciences.

The current version will be supported by the EPFL.

Planned future activities:

  1. extension of the ontology from (mostly) physics to other natural sciences (branches of biology and life sciences; computer science; some branches of humanities; engineering) (work started in collaboration with Swiss Institute for Bioinformatics and other groups)
  2. system is being adapted for learning infrastructure at EPFL (within Learning infrastructure project)
  3. ScienceWISE system is a part of the EPFL's application for "Information Scientifiques" programme (2014 - onward)
  4. collaboration and extension with scientific groups outside physics, with open-access online journals (frontiers), extension of the semantic recommendation system to other domains/services


The ScienceWISE system allows for collecting, storing and semantic searching of scientific data and provides a possibility for a community of scientists, working in a specific domain, to generate dynamically as part of their daily work an interactive semantic environment - field-specific ontology with direct connections to the text of research papers.
The number of registered users continuously grows. Significant user feedback show that by developing several new, highly demanded functionalities, user experience can be improved and the existing platform be used more effectively. Therefore the continuation phase capitalizes on two main elements of the system: its semantic bookmarking and its scientific ontology:

New Arxiv submissions:
develop an improved interface, allowing scientists to specify their "scientific interests" (using the ScienceWISE ontology) and see the daily arXiv submissions reordered accordingly. ScienceWISE system allows to use the existing ontology and users' bookmark collections to find and automatically determine the research interests of a user.

DBpedia--ScienceWISE ontology connection:
adjust the structure of the ScienceWISE ontology with that of DBpedia database to make the relation (export) of DBpedia (and especially of the real-time DBpedia Live content) to the ScienceWISE as seamless as possible.


  • better orientation of young scientists and students in large flow of research papers;
  • attracting increased number of users which additionally ensures sustainability;
  • creation of a new use case to expand and update the ontology.
  • DBpedia project is the largest and highly successful community effort of structuring the information presented in Wikipedia. Synchronizing physics part of the ScienceWISE ontology with the DBpedia will allow not only to harvest the physics pages of Wikipedia in the most efficient way - it will also allow to look up "on the fly" information about the new concepts that users enter into the system.
  • it will help to eliminate a bootstrap when ScienceWISE will be adapted to the non-physics ontologies (eg. Computer Science, Biology, etc.)


Browsing of new articles
The current interface for browsing the new articles at arXiv.org contains no information about the paper apart from its title, authors and subject class(es). Even access to an abstract requires an additional "click". With the growing number of daily submission and quite broad categorization of the arXiv preprints, it becomes more and more difficult for scientists to learn about the new interesting works. The following use-case is planned: Instead a randomly ordered list of papers an interface that reorders these articles according to the user's preferences will be realized.
The default ranking is deduced based on the most relevant concepts from the collection of user's bookmarks (and own papers). In addition there will be a separate interface that would allow scientists to define their interests. The ScienceWISE ontology with its reasoning and relation finding capabilities will allow to make this selection broad and convenient.

At the next stage along with the browsing capability filtering capabilities will be introduced, similar to the filter on the bookmarking page, but based on the daily listing of preprints, allowing users to see at a glance the main topics of the daily submission listings.

DBPedia-ScienceWISE ontology link

  • identify concepts which are the same as in the DBpedia ontology (which contains wikipedia articles brought into a structured RDF format). This will allow to import new relations that are currently present between DBpedia concepts but have no counterparts in the ScienceWISE ontology.
  • use DBpedia templates to infer new information about the existing concepts. This will also allow to look up information about new concepts that scientists enter into the ScienceWISE ontology.
  • export ScienceWISE ontology into the Linked Open Data comparable format allowing to harvest other research databases and in particular export wikipedia concepts and relations more efficiently.