Abstract

SPARQL is a very powerful query language for RDF data, which can be used to retrieve data following specific patterns. In order to foster the availability of scholarly data on the Web, several project and institutions make available Web interfaces to SPARQL endpoints so as to enable a user to search for information in the RDF datasets they expose using SPARQL. However, SPARQL is quite complex to learn, and usually it is fully accessible only to experts in Semantic Web technologies, remaining completely obscure to ordinary Web users. In this paper we introduce OSCAR, the OpenCitations RDF Search Application, which is a user-friendly search platform that can be used to search any RDF triplestore providing a SPARQL endpoint, while hiding the complexities of SPARQL. We present its main features and demonstrate how it can be adapted to work with different SPARQL endpoints containing scholarly data, vis those provided by OpenCitations, ScholarlyData and Wikidata. We conclude by discussing the results of a user testing session that reveal the usability of the OSCAR search interface when employed to access information within the OpenCitations Corpus.

Notes: postprint accepted to SAVE-SD 2018.

This version: https://w3id.org/people/essepuntato/papers/oscar-savesd2018/2018-03-22.html

Last version: https://w3id.org/people/essepuntato/papers/oscar-savesd2018.html

Introduction

The amount of data available on the World Wide Web (the Web) keep increasing rapidly, and finding relevant information by searching the Web is a daily challenge. Traditional search techniques rely on a textual matching of words, and do not take into consideration the semantic information behind the textual content. The Semantic Web is an approach which tries to overcome these disadvantages by representing knowledge on the World Wide Web in a way that can be interpreted by machines. In particular, these data are stored using RDF , a data model that enables one to express information in form of subject-predicate-object statements. Usually these RDF statements are stored in a particular kind of RDF database called a triplestore, and can be queried by means of SPARQL , the query language for RDF data.

SPARQL is a very powerful query language, that can be used to look for data that follow specific patterns. When institutions such as the British Library and the British Museum, and projects such as Wikidata and DBpedia, want to make available their RDF data to the public, they usually provide a specialised Web interface to a SPARQL endpoint of their triplestore, so as to enable users to conduct programmatic searches for particular information, which is returned in one or more formats (usually HTML, XML, JSON, and CSV). However, this SPARQL query language is quite complex to learn, and is normally usable only by experts in Semantic Web technologies, remaining completely obscure to ordinary Web users. The SPARQL service which we have developed for the OpenCitations Corpus is no exception.

The main work of OpenCitations is the creation and current expansion of the OpenCitations Corpus (OCC), an open repository of scholarly citation data made available under a Creative Commons public domain dedication, which provides in RDF accurate citation information (bibliographic references) harvested from the scholarly literature. The whole OCC is available for querying via its SPARQL endpoint, but hitherto it has not had a query interface that would permit ordinary Web users to undertake free text queries in order to to search and explore OCC data.

To address this issue, we have developed an application that can be reused in different contexts for any type of RDF data, not just scholarly RDF data. This is OSCAR, the OpenCitations RDF Search Application, which is a user-friendly search platform that can be used with any RDF triplestore providing a SPARQL endpoint, and which is entirely built without the integration of external application components. OSCAR provides a configurable mechanism that allows one to query a triplestore by means of a user-input free-text string, while in the background one or more SPARQL queries are actually executed. The involvement of experts in Semantic Web technologies remains crucial. However, their involvement is reduced to an initial once-for-all-time configuration of the system to work with a particular triplestore, by customizing the script that provides the text-search interface and enables users to filter the returned results by means of appropriate facets and values.

In this paper, we present the main features of OSCAR and the most significant parts of its configuration file. In order to show how it can be adapted to work with different SPARQL endpoints, we provide three distinct configuration files which implement free-text query interfaces over the SPARQL endpoints of the OpenCitations Corpus , ScholarlyData and Wikidata , respectively. In addition, we report the results of a user testing session in order to understand the perceived usability of one of the aforementioned implementations, namely that for searching the OpenCitations Corpus.

The rest of this paper is organized as follows. In we describe some of the most important existing SPARQL-based searching tools. In , we describe OSCAR and discuss its model definition and architectural form. In , we demonstrate its reusability in different contexts, while, in , we assess the usability of the search interface of this application when querying the OpenCitations Corpus. Finally, in , we conclude the paper and sketch out some future developments.

Related works

In the past, several projects that customize SPARQL queries according to user needs have been released. They can be classified into two categories.

On the one hand, there are the tools that generate and apply SPARQL queries starting from a free text search input, hiding the complexities of the SPARQL query behind a simple and familiar-looking search interface. We say that these interfaces are unaware-user tools, since they permit users to make textual queries without needing to understand the complexity of the languages used for storing and querying the data.

Scholia is a tool which is in this category. It is a Web application to expose scientific bibliographic information through Wikidata. In particular, its Web service creates on-the-fly scholarly profiles for researchers, organizations, journals, publishers, individual scholarly works, and research topics, by querying the SPARQL-based Wikidata Query Service. A search field on the front page of Wikidata permits a user to look for a particular name and displays its data by means of well-structured visual interfaces.

Another tool of this type is BioCarian . It is an efficient and user-friendly search engine for performing exploratory searches on biological databases, providing an interface for SPARQL queries over RDF triplestores, and providing a graphical interface for the results based on facets. It allows complex queries to be constructed, and has additional features like filtering and ranking the results according to user-selected facet values.

OSCAR is, of course, also in this category of tools.

On the other hand, there are tools which aim at helping the user to build a SPARQL query by using specific visual constructs that mimic the various operations made available by SPARQL (filters, values selections, etc.). In this case, the users are very aware that they are using Semantic Web technologies, and that the function of the tools is only to be a support for guiding the user in creating the particular query of interest. These tools are grouped under the label aware-user tools.

Within this category, we have the Wikidata Query Service (WDQS), which is a Web interface that enable the creation of a SPARQL query by writing the actual label associated to each item and property, instead of employing the URL customarily used for identifying it. In addition, it makes available several visualisation interfaces for the results, from a simple table to very complex diagrams and graphs.

Along the same lines is the Visual SPARQL Builder (VSB), a tool which allows users to create and run SPARQL queries with a browser's graphical interface . While WDQS exposes the SPARQL query to the user, VSB hides it entirely behind blocks which explicitly represent all the SPARQL constructs, including filters. Upon execution, the visual queries are translated to SPARQL code, which is made accessible to the user, and the results are shown in structured tables.

OSCAR, the OpenCitations RDF Search Application

OSCAR, the OpenCitations RDF Search Application, is an open source stand-alone javascript tool which can be embedded in a webpage so as to provide a human-friendly interface for searching for data within RDF triplestores by means of SPARQL queries. It is possible to configure OSCAR to work with a particular SPARQL endpoint, by configuring a JSON document which specifies how the SPARQL queries are sent to that endpoint, and how the returned query results should be visualized, according to the predefined tabular view that OSCAR provides. The source code and documentation for OSCAR are available on GitHub at https://github.com/opencitations/oscar.

OSCAR has been developed to meet the following requirements:

• it must enable a free text search, such as is common to Web search engines;

• it must permit filtering of the result set, filtering operations being applied to one or more of the result fields presented in the tabular results interface, and these operations must be applied dynamically and handled without the need for any further querying of the triplestore;

• its interface, functionalities and queries must be customizable according to the user needs;

• although originally developed to work with the OpenCitations SPARQL endpoint, it must be easily configured to work with any other RDF triplestore, and must also be easily integrated as a new module within that system's website.

The above list of requirements is based on our observations and experience of user requirements while working on the OpenCitations project, and on informal feedbacks from potential users interviewed during OSCAR's development. In the following subsections we describe the general architecture of the OSCAR system.

               
<div id="search" class="search">
  <div id="search_extra" class="search-extra"></div>
  <div id="search_header" class="search-header">
    <div id="rows_per_page"></div>
    <div id="sort_results"></div>
  </div>
  <div id="search_body" class="search-body">
    <div id="search_filters" class="search-filters">
      <div id="limit_results"></div>
      <div id="filter_btns"></div>
      <div id="filter_values_list"></div>
    </div>
    <div id="search_results" class="search-results"></div>
  </div>
</div>

            

               
{
  "sparql_endpoint": "https://w3id.org/oc/sparql",

  "prefixes": [
    { "prefix":"cito", "iri":"http://purl.org/spar/cito/" },
    { "prefix":"dcterms", "iri":"http://purl.org/dc/terms/" },
    ...
   ],

  "rules": [
    {
      "name":"doi",
      "category": "document",
      "regex":"(10.\\d{4,9}\/[-._;()/:A-Za-z0-9][^\\s]+)",
      "query": "SELECT DISTINCT ?iri ?short_iri ..."
    },
    ...
  ],

  "categories": [
    {
      "name": "document",
      "fields": [
        {
          "value":"short_iri", "title": "Corpus ID", 
          "column_width":"15%", "type": "text", 
          "sort": { "value": true }, 
          "link": { "field":"iri", "prefix":"" }
        },
        ...
      ]
    },
    ...
  ]
}
            

Configuring OSCAR for use with different RDF datasets

In order to demonstrate the flexibility that OSCAR offers in terms of its customizability and adaptability to a particular triplestore, we have developed three configuration files that provide free-text search interfaces to three different RDF datasets (each with a separate SPARQL endpoint): those of OpenCitations Corpus, Scholarly Data and Wikidata.

The OpenCitations Corpus (OCC, http://opencitations.net) is an open repository of scholarly citation data bibliographic information that we have developed, which has been the main target and incentive for the development of OSCAR. The OCC triplestore contains such open scholarly citation and bibliographic data. The OSCAR search interface for the OpenCitations Corpus can be seen and used at http://opencitations.net/search. The use of OSCAR in this context enables the search for two basic entities included in the OCC: documents (bibliographic resources) and authors. Currently, the free-text search allows the recognition of two different types of input: unique global identifiers (specifically DOIs and ORCIDs, that identify published works and people, respectively), and any other textual string which can be used to identify the title of a document or the name of an author. It is worth mentioning that this text search string do not match against the abstracts and the keywords of documents since these data are not currently stored within the OCC. In , we depict a screenshot of OSCAR after the execution of a free text search using the string machine learning.

A similar instantiation has been made for the SPARQL endpoint provided by Scholarly Data (http://www.scholarlydata.org) . Scholarly Data provides an RDF dataset of the papers, people, organisations and events related to Web and Semantic Web academic conferences. The current version of the project has a SPARQL endpoint, but it does not include a free-text search interface for discovery of the entities included in the dataset. We have created an ad hoc configuration file for OSCAR so as to enable such search operation on the Scholarly Data triplestore. The Web interface for OSCAR searches over the SPARQL endpoint for Scholarly Data is available at http://opencitations.net/static/savesd2018/oscar/scholarlydata.html. shows a screenshot of OSCAR with ScholarlyData results.

Finally, we have prepared a configuration files to enable free-text searches within Wikidata (https://www.wikidata.org) . Wikidata is a free open knowledge base which acts as a central store for the structured data of Wikimedia Foundation projects including Wikipedia, and of other sites and services. Wikidata offer a SPARQL query service and already has its own powerful Web graphical user interface for facilitating the users to construct SPARQL queries, Scholia, as mentioned above. Our OSCAR interface to the Wikidata SPARQL endpoint is thus entirely for demonstration purposes, rather than to provide new functionality for Wikidata users. While Wikidata contains a wide variety of information, for the Wikidata customisation of OSCAR we decided to limit the range of data to be searched to bibliographic entities within the scholarly domain. The configuration file thus includes rules for detecting DOIs, ORCIDs and free textual inputs, and classifies the information returned according to two categories: documents and authors, as is done for the OSCAR search interface over the OpenCitations Corpus. The Web interface for OSCAR searches over the SPARQL endpoint of Wikidata is available at http://opencitations.net/static/savesd2018/oscar/wikidata.html. shows a screenshot of OSCAR with Wikidata results.

Empirical evaluation

In order to measure the perceived usability of the OSCAR search interface over the OpenCitations Corpus data, we involved five subjects in a user testing session. The test was organized as a Discount Usability Test (DUT): a low-cost test requiring few participants. Past studies have demonstrated that even when involving only five people, such a test can find up to 80% of the main problems of a system. In the next subsections, we first discuss the settings of our tests and then we show the outcomes. All the material and outcomes of the test are available at .

Conclusions

In this paper we have introduced OSCAR, the OpenCitations RDF Search Application, a user-friendly searching tool for use with RDF triplestores presenting a SPARQL endpoint. In particular, we have presented its main features and we have shown how it can be adapted to work with different SPARQL endpoints, including those of the OpenCitations Corpus, Scholarly Data and Wikidata. Finally, we have discussed the outcomes of user testing undertaken in order to understand the perceived usability of the OSCAR search interface when used to interrogate the OpenCitations Corpus.

We are currently working on some extensions of OSCAR, particularly an advanced search interface, so as to permit complex combinations of field-oriented queries by means of logical connections (OR, AND, etc.). For a more accurate usability analysis, we would like to strengthen our evaluation by involving a larger number of participants from different fields of study, and also to evaluate OSCAR in a comparative test with other similar tools. In addition, we are presently developing a related tool, named LUCINDA, to browse the resources inside a triplestore. In particular, LUCINDA will have the ability to visualize the resources attributes directly, providing detailed information about bibliographic resources such as journal names, page numbers, and additional identifiers. As with OSCAR, this new browsing tool will be integrated within the OpenCitations Web site so as to provide human-readable descriptions of the OpenCitations Corpus entities, and will equally be made publicly available for customization and implementation over the triplestores of other RDF data repositories.

We recommend the use of OSCAR, which is a fully open source application provided by OpenCitations for general community use, to permit free-text queries over the SPARQL endpoints of other triplestores. We would be grateful to receive feedback (to contact@opencitations.net) from its users.

Acknowledgements

We gratefully acknowledge the financial support provided to us by the Alfred P. Sloan Foundation for the OpenCitations Enhancement Project (grant number G‐2017‐9800).

References

1. Nielsen, F., Mietchen, D., Willighagen, E. (2017). Scholia, scientometrics and Wikidata. In Proceedings of the Satellite Events of the 14^th Extended Semantic Web Conference (ESWC 2017): 237-259. DOI: https://doi.org/10.1007/978-3-319-70407-4_36

2. Silvio Peroni, Alexander Dutton, Tanya Gray, David Shotton (2015). Setting our bibliographic references free: towards open citation data. Journal of Documentation, 71 (2): 253-277. DOI: http://dx.doi.org/10.1108/JD-12-2013-0166; OA at http://speroni.web.cs.unibo.it/publications/peroni-2015-setting-bibliographic-references.pdf

3. Nuzzolese, A. G., Gentile, A. L., Presutti, V., Gangemi, A. (2016). Conference Linked Data: The ScholarlyData Project. In: Proceedings of the 15th International Semantic Web Conference (ISWC 2016): 150-158. DOI: https://doi.org/10.1007/978-3-319-46547-0_16

4. Zaki, N., Tennakoon, C. (2017). BioCarian: search engine for exploratory searches in heterogeneous biological databases. BMC Bioinformatics 18: 435. DOI: https://doi.org/10.1186/s12859-017-1840-4

5. Nielsen, J. (2000). Why you only need to test with 5 users. https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/ (last visited 30 January 2018)

6. Nielsen, J. (2009). Discount usability: 20 years. https://www.nngroup.com/articles/discount-usability-20-years/ (last visited 30 January 2018)

7. Sauro, J. (2011). A Practical Guide to the System Usability Scale: Background, Benchmarks & Best Practices. ISBN: 978-1461062707

8. Cyganiak, R., Wood, D., Lanthaler, M. (2014). RDF 1.1 Concepts and abstract syntax. W3C Recommendation 25 February 2014. https://www.w3.org/TR/rdf11-concepts/

9. Harris, S., Seaborne, A. (2013). SPARQL 1.1 Query Language. W3C Recommendation 21 March 2013. https://www.w3.org/TR/sparql11-query/

10. Vrandečić, D., Krötzsch, M. (2014). Wikidata: a free collaborative knowledgebase. Communications of the ACM, 57 (10): 78-85. DOI: http://doi.acm.org/10.1145/2629489

11. Peroni, S., Shotton, D., Vitali, F. (2017). One year of the OpenCitations Corpus - releasing RDF-based scholarly citation data into the public domain. In Proceedings of the 16^th International Semantic Web Conference (ISWC 2017): 184-192. DOI: https://doi.org/10.1007/978-3-319-68204-4_19

12. Hiebi, I., Peroni, S., Shotton, D. (2018). Material and outcomes of the OSCAR user testing session. Figshare. DOI: https://doi.org/10.6084/m9.figshare.5849724

13. Lewis, J. R., Sauro, J. (2009). The factor structure of the System Usability Scale. In Proceedings of the 1st International Conference on Human Centered Design (HCD09). DOI: https://doi.org/10.1007/978-3-642-02806-9_12

14. Strauss, A., Corbin, J. (1998). Basics of Qualitative Research Techniques and Procedures for Developing Grounded Theory (2nd edition). Sage Publications: London. ISBN: 978-0803959408