Method and device for evaluating and supporting improvement of web accessibility by using generative ai

Abstract

To provide a method and a device for evaluating and supporting improvement of web accessibility by using a generative AI, the method and the device allowing a web developer to get real-time feedbacks and propositions of improvement.SOLUTION: In a system for evaluating and supporting improvement of web accessibility by using a generative AI, a server 101 accesses the URLs of web sites 601, 602 and evaluates an access barrier in accessibility to the acquired source code, and a generative AI 201 creates a specific proposition of improvement on the basis of the result of the evaluation.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to the evaluation and improvement of web accessibility, and particularly belongs to the technical field of automatically identifying web accessibility problems of websites by utilizing machine learning and generating practical improvement suggestions. This technology aims to improve the ease of web access and focuses on enabling all users, including those with disabilities, to easily access web content.

Background Art

[0002] In recent years, with the popularization of the Internet, web accessibility has become an important social issue. Web accessibility refers to providing a website with inclusive design that can be used by all people regardless of disabilities or differences such as age. In Japan, the accessibility standards that web content should meet are defined in Japanese Industrial Standard JIS X8341-3. In addition, with the amendment of the Act on Elimination of Discrimination against Persons with Disabilities (Act on Promotion of Elimination of Discrimination on the Grounds of Disability) in 2021, in addition to national and local public bodies, private businesses will also be obliged to respond to web accessibility from fiscal year 2024.

[0003] However, many websites do not fully meet the accessibility requirements for users with disabilities or the elderly. Conventional web accessibility evaluation methods often rely on manual checks based on checklists, which are time-consuming and labor-intensive tasks. In addition, there are situations where there is a lack of guidance for finding improvement measures after identifying problems. Many web developers do not have sufficient expertise in web accessibility, and it is extremely difficult to take sufficient measures regarding web accessibility in the above situations.

[0004] Under such a background, the present invention provides a new technology for automating the accessibility evaluation of websites and efficiently making improvement suggestions. In particular, an automatic evaluation and improvement suggestion generation system using machine learning technology can quickly evaluate the compliance of websites and point out specific improvement points, enabling web developers to easily achieve improvements in accessibility. Furthermore, the present invention aims to continuously improve the quality of improvement suggestions by incorporating user feedback to evolve the machine learning model. This approach provides a practical and effective solution for solving web accessibility problems and improves the usability of websites for all users.

Summary of the Invention

Problems to be Solved by the Invention

[0005] The purpose of the present invention is to address specific technical problems by automating the process of web accessibility evaluation and improvement and providing quick and efficient improvement suggestions. In existing web accessibility evaluation methods, in addition to the need for expertise across many technical elements to conduct comprehensive evaluation tests, manual evaluation is time-consuming and prone to subjective errors. Also, due to the insufficient means for generating executable improvement plans after evaluation, it has been difficult for web developers to effectively solve accessibility problems. This has led to a situation where users with restricted access, such as disabled users and the elderly, are limited in enjoying equal access to information.

[0006] According to the present invention, it is possible to improve the accuracy and speed of automatic evaluation of web accessibility and improvement proposals. This technology uses a machine learning model to evaluate the code of a website and indicates specific improvement points and directions for improvement for accessible web design. Furthermore, code snippets for fixing problems are recommended. As a result, web developers can quickly improve the accessibility of their sites. Furthermore, the machine learning model can evolve by performing machine learning using feedback from users and generate more accurate improvement proposals. Therefore, the present invention provides a more effective and practical solution as it is more widely used in the field of web accessibility.

Means for Solving the Problems

[0007] Embodiments of the present invention relate to an automated method and apparatus for identifying and solving web accessibility problems. As a series of processes for evaluating the accessibility of a website and assisting in its improvement, a reception stage of accessing the URL of a specified website and obtaining source code such as HTML, CSS, and JavaScript of the website, an analysis stage of performing an analysis on the obtained source code based on pre-defined accessibility criteria, an improvement proposal generation stage of identifying elements with problems based on the analysis results and generating improvement proposals indicating specific improvement measures, and a transmission stage of presenting the generated improvement proposals to the user.

[0008] In addition to the means of accessing the URL of the website and obtaining the source code, the reception stage may include means of directly obtaining the source code to be analyzed from a web developer who is a user of the present invention.

[0009] The analysis in the analysis stage may include a plurality of checkpoints based on the accessibility criteria, such as compatibility with screen readers, color contrast ratios, accessibility of links, and labeling of form controls.

[0010] The transmission stage is often carried out through a dashboard for web developers, but it may also be integrated into other systems through an API.

[0011] The implementation device of the present invention consists of hardware and software components for automating the aforementioned method, and includes a receiving module that accesses a specified URL and obtains code and configuration files from the target website, and an AI module (hereinafter referred to as the "accessibility evaluation AI module") that functions by a machine learning model that performs an accessibility check on the obtained code and generates an evaluation result, and an AI module (hereinafter referred to as the "improvement proposal AI module") that functions by a machine learning model that performs improvement proposal generation to generate one or more improvement measures for the problems identified based on the evaluation result, and a user interface module that presents the improvement proposals to the user in an easy-to-understand manner and collects feedback.

[0012] According to the present invention, by automatically evaluating web accessibility and generating improvement proposals, the speed and accuracy of improving the web accessibility of websites are greatly improved. In particular, the present invention is characterized in that it uses a machine learning algorithm to improve the quality of the provided improvement proposals over time, and web developers can respond quickly and effectively based on the provided improvement proposals. In addition, the present invention eliminates subjective judgments when evaluating the accessibility of websites and standardizes the evaluation and improvement processes.

Advantages of the Invention

[0013] The present invention has the following multiple remarkable effects by efficiently and effectively identifying and improving web accessibility problems.

[0014] First, since the present invention automates the accessibility evaluation of websites and the generation of improvement proposals, the time and effort required for the process from website accessibility evaluation to improvement are significantly reduced. Compared with the conventional manual evaluation process, a quick and accurate evaluation becomes possible, and specific improvement proposals based on the evaluation results are automatically generated, so web developers can obtain real-time feedback and improvement proposals.

[0015] Next, by promoting machine learning using machine learning algorithms, the accuracy of improvement proposals regarding website accessibility is improved, providing more specific and executable solutions. As a result, web developers can quickly identify problems and easily make more effective corrections.

[0016] Furthermore, the present invention can generate customized improvement proposals that go beyond general guidelines regarding web accessibility and are tailored to the specific context of the website and the specific needs of users. This substantially helps all users to equally access web content.

[0017] Finally, the web accessibility evaluation and improvement support system provided by the present invention promotes compliance with laws and regulations and serves as a means for companies and organizations to fulfill their social responsibilities. By using this system, it contributes to promoting an inclusive design of the web and realizing a more accessible digital environment for all users.

Brief Description of the Drawings

[0018] The drawings cited from the detailed description of the present invention will be described.

[0019]

Figure 1

Figure 2

Figure 3

Figure 4

Mode for Carrying Out the Invention

[0020] The present invention relates to a method and system for evaluating and improving web accessibility, and particularly provides an automated solution for discovering and improving web accessibility barriers of a website by applying machine learning.

[0021] FIG. 1 is a block diagram of a web accessibility evaluation / improvement support system according to the present invention.

[0022] As shown in FIG. 1, the system of the present invention includes a server (element number 101), a generation AI (element number 201), a reception module (element number 301), a user interface module (element number 401), and client devices (element numbers 501, 502,...).

[0023] The generation AI (element number 201) is an AI module that functions by a machine learning model, and includes an accessibility evaluation AI module (element number 202) that performs an automatic evaluation of accessibility, and an improvement proposal AI module (element number 203) that generates improvement proposals based on the evaluation results.

[0024] In the present embodiment, as shown in FIG. 1, the web accessibility evaluation and improvement support system is composed of a series of modules including an external generation AI (element number 201), and the generation AI is connected and operates in cooperation with the server (element number 101) via a network.

[0025] The generation AI (element number 201) shown in FIG. 1, the accessibility evaluation AI module (element number 202) and the improvement proposal AI module (element number 203) included therein may be accommodated in the server if the server (element number 101) has sufficient processing power.

[0026] The reception module (element number 301) in FIG. 1 accesses the URLs of websites (element numbers 601, 602,...) and acquires the source code of web pages (element numbers 601-1, 601-2,...) included in the websites and configuration files such as images.

[0027] The reception module (element number 301) can also directly acquire the source code and configuration files of a web page from a user via the user interface module (element number 401). Thereby, for example, a web page that has not yet been publicly available on the Internet can be set as an evaluation target.

[0028] The accessibility evaluation AI module (element number 202) included in the generation AI (element number 201) in FIG. 1 analyzes the source code acquired by the reception module, performs an evaluation based on the accessibility evaluation criteria, and identifies problems.

[0029] Based on the evaluation results derived by the accessibility evaluation AI module, if there are identified problems, the improvement proposal AI module (element number 203) in FIG. 1 generates at least one code correction or design change proposal to address the problems.

[0030] The user interface module (element number 401) in FIG. 1 presents the evaluation results derived by the accessibility evaluation AI module and the improvement proposals generated by the improvement proposal AI module for the identified problems to the web developer who is the user. This is realized through the visual representation of the evaluation results and improvement proposals as shown in FIGS. 2 and 3, and the web developer responds based on these improvement proposals.

[0031] FIG. 2 schematically shows an example of the evaluation results presented by the user interface module to the user. In the example of this figure, the web page shown at ▲2▼ is evaluated, the evaluation result summary shown at ▲1▼ is output, and the problems identified at ▲3▼ are shown. It is an example of the display showing the relationship between the display image of the web page, its source code, and the identified problems. Through such a display, the web developer who is the user of this system can visually grasp the location and content of the problems regarding web accessibility.

[0032] FIG. 3 schematically shows an example of the improvement proposals presented by the user interface module to the user. In the example of this figure, the outline of the improvement proposal is shown at ▲4▼, and two improvement proposals and their code snippets are presented at ▲5▼. Note that in this example, the improvement proposals are embodied as code snippets, but depending on the content of the identified problems, a more extensive corrected source code may also be presented as an improvement proposal.

[0033] In the example shown in FIG. 3, a link for obtaining feedback from the user is displayed at <6>. In the example of this figure, the user provides feedback information on which improvement plan to adopt to the generative AI by selecting any of the links at <6>. Instead of a link, an input form may be displayed at <6> so that the user can input feedback information in free text.

[0034] FIG. 4 is a flowchart for explaining the basic operation of the system shown in FIG. 1.

[0035] In the flowchart shown in FIG. 4, at step number S102, the receiving module accesses the URL of the website, and at step number S103, it obtains the source code of the website. Subsequently, the analysis of the obtained source code by the generative AI starts at step number S104. It shows the process where the generative AI identifies the web accessibility issues included in the website at step number S105 and creates specific improvement proposals at step number S106. The web developer, who is the user of this system, can receive the visualized evaluation results and improvement proposals as shown in the examples of FIGS. 2 and 3 at step number S107 and take measures to effectively and quickly eliminate the access barriers.

[0036] In addition to the above series of operations, feedback from the user is collected at step number S108 in FIG. 4, and the generative AI uses this to continuously improve its machine learning model by machine learning at step number S109 in FIG. 4. This process is essential for the system to adapt to various websites and web user conditions and be able to provide more appropriate accessibility improvement proposals.

[0037] Embodiments of the present invention provide a comprehensive framework for automated web accessibility assessment and improvement, enabling web developers without specialized knowledge of web accessibility to make their web sites more accessible, facilitating legal compliance in web site provisioning, and improving web access for all users.

Claims

1. A method for evaluating the accessibility of a website using one or more machine learning models and generating improvement suggestions based on the evaluation results, comprising the following steps: (a) Obtaining the source code of the website by means such as accessing the URL of the website; (b) Analyzing the obtained website code according to accessibility evaluation criteria; (c) Generating improvement suggestions for instructing improvement of accessibility based on the analysis; (d) Presenting to the user an improvement plan for the website using the improvement suggestions.

2. The method according to claim 1, wherein the machine learning model continuously learns and improves by receiving feedback from users.

3. The method according to claim 1 or 2, wherein the improvement suggestions for instructing improvement of accessibility present multiple solutions for specific accessibility problems.

4. An apparatus for evaluating web accessibility and generating improvement suggestions, comprising the following modules: (a) A receiving module; (b) An AI module that functions by a machine learning model for performing accessibility evaluation; (c) An AI module that functions by a machine learning model for performing improvement suggestion generation; (d) A user interface module.

5. The apparatus according to claim 4, wherein the machine learning model utilizes data collected via a user interface to generate improvement suggestions for instructing improvement of accessibility.

Images