W3C homepageWeb Accessibility initiative homepage

WAI: Strategies, guidelines, resources to make the Web accessible to people with disabilities

3 December 2014

WAI R&D Symposia » Metrics Home » Proceedings » This paper.

This paper is a contribution to the Website Accessibility Metrics Symposium. It was not developed by the W3C Web Accessibility Initiative (WAI) and does not necessarily represent the consensus view of W3C staff, participants, or members.

Extended Abstract for the RDWG Symposium on Accessible Way-Finding Using Web Technologies

Applying the Linked Data Principles to Open Accessibility Data for Accessible Way-Finding

  • Chaohai Ding, University of Southampton, cd8e10@ecs.soton.ac.uk.
  • Mike Wald, University of Southampton, mw@ecs.soton.ac.uk.
  • Gary Wills, University of Southampton, gbw@ecs.soton.ac.uk.

1. Problem Addressed

There are many projects attempting to address the accessibility challenges of day-to-day activities for people with disabilities or special needs. Accessible way-finding for people with disabilities is a significant research area to improve urban accessibility. One of the main challenges is the lack of useful accessibility information for physical places and objects. Another challenge is that the accessibility information data is difficult to reuse and isolated from different systems, projects or applications, which means that it cost’s money and is time consuming for people with disabilities to find such accessibility information to prepare their trip or search during the trip. Therefore, in this paper, we present the term of open accessibility data and the approach to use the Linked Data principles to integrate various accessibility data, thereby improving data sources for accessible way-finding.

2. Background

There is no standard definition for open accessibility data. In our research, the term open accessibility data is the open data of physical places with additional accessibility information, such as step-free access, accessible entrance, accessible toilets, accessible car parking, braille menu or hearing loops etc. Open accessibility data also refers to the data that benefits people with special needs, such as baby change facilities, staff help available points, customer help points, whether travelling with a baby pushchair or large luggage etc. The term open accessibility data is also already applied in several projects. For example, the accessibility information of London tube stations used in RailGB [1], the smart data presented in mPass project [2], and shared annotation of accessibility information for surroundings in the OurWay project [3]. Based on the survey of open accessibility data in our previous research[4], there are some issues summarized as follows:

  • There are some datasets with accessibility information available online, which refer to accessibility information of physical places and objects.
  • Open accessibility data is from multiple heterogeneous resources with various formats. Some datasets are well structured and presented in standard formats (e.g., XML, JSON and CSV etc.), while some are not in standard open data formats (e.g., HTML, PDF or even in image formats etc.).
  • There is no standard guideline to specify accessibility information for physical places or objects due to the domain or requirements differences between various applications.
  • Crowdsourcing is powerful but limited by too much incomplete information. The quality of open accessibility data published by the government or organizations is better than the crowdsourced data.

3. Strategy

Therefore, in order to address these challenges: the lack of accessibility information and the data isolated from different systems or projects, we proposed the approach to apply the Linked Data principles to publish, interlink and thereby integrate the open accessibility data from heterogeneous sources. As the core data layer in the Semantic Web, the basic idea of Linked Data provides the principles to create and publish structured and machine-readable data with additional semantic linking to other resources on the Web. The data could be linked to other resources from external datasets and thereby expose the significant advantages, such as human and machine readable data, well-structured standard data format, domain specified, semantic linking and openness [5]. The main strategy for publishing open accessibility data as the Linked Data is following the four rules of Linked Data principles and links to other resources to achieve the 5-star Linked Data [6].

4. Major Difficulties

One of major difficulties is the lack of standard ontologies for modelling and describing the accessible places and objects. This results from the fact that there is no standard guideline to specify accessibility information for physical places or objects while another major difficulty is the approach for data integration. There are three different ways to integrate data based on ontology, namely the single ontology approach, multiple ontologies approach and hybrid approach. However, using the single ontology approach is faced with the problems of the mapping between low dimensional data (only one attribute to describe accessibility) and high dimensional data (multiple attributes to describe accessibility). The hybrid approach is faced with the problem to design the ontologies for different applications. The multiple ontologies approach is faced with the problem of mapping between the multiple ontologies, which is still a challenging issue and involves a large number of research efforts.

5. Outcomes

In our recent work [7], we developed the placeaccess ontology and applied the single ontology approach as well as matching rules to publish the four open accessibility datasets as Linked Data. We also implemented the accessibility data crowdsourcing platform to improve accessibility data quality and interlink the same entities from different datasets. Table 1 demonstrates the integration result of the open accessibility data extracted from four different systems, namely Wheelmap , National Rail , London Tube stations, and Factual restaurants.

Table 1: Open Accessibility Data Integration Result
Train Station(Wheelmap) Train Station(National Rail) Tube Station(Wheelmap) Tube Station(Transport for London) Food(Wheelmap) Restaurant(Factual)
Total 3,384 2,601 222 362 56,970 10,629
Entity with Name 3,323 2,601 119 362 52,901 10,629
Entity without Name 61 0 103 0 4,071 0
Total Matched Entities 2,415 (92.85%) 48 (21.62%) 1,835 (17.27%)
Entities Enrichment 2,216 (92.76%) 20 (41.67%) 1,724 (93.95%)
No Change 164 (6.79%) 26 (54.17%) 60 (3.27%)
Conflict 35 (1.45%) 2 (4.16%) 51(2.78%)

According to the result, there are 92.85% of entities in National Rail dataset matched with the train stations in Wheelmap. Only 21.62% of tube stations are matched with the tube stations in Wheelmap, while 17.27% of the restaurants in Factual are matched with restaurants in Wheelmap. There are 35 entities that conflict in the train station matching. There are 2 entities conflicted in tube station datasets matching and 51 entities conflicted in restaurant datasets mapping. However, there are also many entities that are enriched with accessibility information. For example, there are 2216 train stations (92.76%), 20 tube stations (41.67%) and 1724 restaurants (93.95%) matched to the equivalent entities in Wheelmap dataset. Moreover, the entities are also links to the same entities in the LinkedGeoData and there is a SPARQL endpoint published for data querying, updating and interlinking.

6. Open Research Avenues

The major difficulties mentioned in this paper are also the emerging challenges to address, such as the standard ontology for annotating the accessibility information of physical places and objects, the integration approaches, and accessibility data conflict after the open accessibility data integration. There are many research avenues closely related to this project. Firstly, this approach could address the lack of accessibility information for the physical places and objects. It could also address the problem of data isolation by applying Linked Data principles to establish a public linked open accessibility data integrated from heterogeneous data sources. It also makes possible research into optimized accessible way-finding based on the linked open accessibility data.

References

  1. Li, Y., Draffan, E.A., Glaser, H., et al. RailGB : Using Open Accessibility Data to Help People with Disabilities. Proceedings of the Semantic Web Challenge co-located with ISWC2012, (2012), 1–8.
  2. Prandi, C. Accessibility and Smart Data : the Case Study of mPASS. Proceedings of 13th International Web for All Conference - W4A ’14, (2014), 9–10.
  3. Holone, H. and Misund, G. People helping computers helping people: Navigation for people with mobility problems by sharing accessibility annotations. Computers Helping People with Special Needs, (2008).
  4. Ding, C., Wald, M., and Wills, G. A Survey of Open Accessibility Data. Proceedings of 13th International Web for All Conference - W4A ’14, ACM (2014), 73–80.
  5. Bizer, C., Heath, T., and Berners-Lee, T. Linked data-the story so far. International Journal on Semantic Web and Information Systems (IJSWIS) 5, 3 (2009), 1–22.
  6. Tim Berners-Lee. Linked Data - Design Issues. 2009. http://www.w3.org/DesignIssues/LinkedData.html.
  7. Ding, C., Wald, M., and Wills, G. Open Accessibility Data Interlinking. Proceedings of The 14th International Conference on Computers Helping People with Special Needs (ICCHP), (2014), 73–80.