For people with disabilities, finding jobs with understanding, accommodating employers can be a challenge. To help facilitate these opportunities, one company provides a website that connects employers with people with disabilities. The process is simple: candidates can search for positions and employers can search for candidates, but the system itself does not attempt to make these connections on its own.
To achieve this type of service, a human (for example, somebody working in vocational rehab) would have to review the candidates and positions on the site to match them up manually. In an effort to expand and automate these services, the company partnered with the Microsoft AI for Accessibility program and BlueGranite to develop an AI-driven solution with an emphasis on accessibility.
The company had an existing source of data from employers, but it was largely unstructured with no metadata regarding what type of position they were offering or what the position’s requirements were. The existing data about candidates consisted of a resume-style job description for any previous work experience.
In addition to structuring this data, there was also a desire to keep the interface between human and machine as accessible as possible. Accessibility accommodations on the site needed to support physical disabilities such as impaired sight or limited limb movement and cognitive disabilities such as high anxiety, autism spectrum disorder, or ADHD.
To assist job seekers without any prior experience or planned career path, a questionnaire encompassing basic skills was developed by the company's team. Responses to these questions were then mapped to frequently occurring broad skills. To facilitate even deeper skill matches, BlueGranite built a program using Azure Cognitive Services Text Analytics API that analyzed text from applicants' previous job descriptions and employers' current job descriptions to pick out skills from these bodies of text.
By understanding an individual's skills and comparing this set of skills against all open positions, the company could now make targeted recommendations as to which jobs might be the right fit.
From the summation of skills identified in each job and a candidate’s identified skills, a ranked list of positions could then be computed and returned from the AI system. Then, this result could be filtered down to positions that are available within a certain radius based on geocoded data obtained from Azure Maps. A future enhancement to this system will allow for filtering based on commute time or restrictions on what positions are available by public.
To support users with cognitive disabilities, the entire interface to the profile-building process was constructed in a chatbot. This chatbot is embedded into the website as an assistant built using the Microsoft Bot Framework. This provides an interface to the job seeker that is single focus in nature, can be stopped or resumed at any time, and never encumbers the job seeker with more than a handful of questions (8-10) in a row.
In addition, the Microsoft provided a webchat interface that includes excellent text-to-speech and speech-to-text services for users who might benefit from them but might not have screen reader software available at the time.
Establishing a successful AI strategy for a business requires more than sophisticated tools and data scientists - it starts with building a strategy.
If you'd like to experience the transformational power of machine learning and AI, contact BlueGranite today!