Objectives

The aim of the OAK project is to design and implement a proof-of-concept (POC) that will enable a skilled operator’s practical experience to be collected, stored and explored in a simple, non-invasive way, in the form of a Knowledge Base (KB), transforming it into “Actionable Knowledge”. Actionable Knowledge” refers to the knowledge that actors can use to effectively implement their intentions. This tool will help operators in their day-to-day activities, simplify the sharing of tasks between teams working in rotation, and simplify training for the next generation.

Method

User-centered and agile design approaches will be used to implement the two main parts of OAK Tutor: the multimodal interface and data analysis.

For the optimization of the data analysis and knowledge representation part, knowledge graph embedding approaches will be used. In the context of this project, this implies that we will test several approaches in order to find the most suitable for our problem and our data. Based on the experience of the research team and the state of the art, we will focus on approaches based on knowledge graph embedding, which combine machine learning approaches with the possibility of superimposing a more formal information structure. In concrete terms, this will enable us to meet the needs of companies used to using skills matrices for human resources management, for example.

The platform comprises the following modules:

• OAK Tutor – Multimodal interface: This component manages interaction with users and will be designed to support them in the tasks of recording and annotating new information as well as searching for existing information.
• OAK Tutor – Data Analytics: This component contains all the elements needed to dynamically create a structured knowledge base from multimedia elements.

Impact

a.      Scientific impact

A first key innovative element of the project is the exploitation of big data techniques for small quantities of data. This “small data” represents 85% of all data collected by companies, and is a key innovation vector for the future of small industries. The aim of the project is to use and adapt “big data” approaches (data analysis and processing) in an intelligent way to efficiently process the data collected by employees in their day-to-day work, and improve the decision-making process. The second innovative element is the design of intuitive multimodal interactions. Exploring new modes of interaction that promote natural interaction integrated into daily routines is a key element in supporting operators in their work and facilitating the transmission of knowledge. The knowledge developed as part of this project will form a solid basis that can be transposed to other contexts for learning professional practices.

b.      Technological impact

To assess a situation and choose a solution, an experienced operator mobilizes knowledge or experience resources that may go beyond the scope of standard situational knowledge. Our demonstrator will enable us to capitalize on the lessons learned from experience and professional practice, and facilitate their collection and sharing within a company (actionable data). The originality of this demonstrator lies in the implementation of a simple tool, which integrates seamlessly into the company’s current daily work routines, because: 1) with the same tool, operators can both create and feed the company’s knowledge base, but also access information from the base when needed, 2) information gathering is done by exploiting the principles of natural interaction and automation (voice recognition for automatic and contextual note-taking, thanks to the use of object recognition techniques).

c.       Socio-economic impact

The main aim of the project is to support companies’ digitalization process by helping them to capitalize on and perpetuate the knowledge and experience of experienced operators and employees, for the benefit of colleagues (job succession) and novices (succession training). On a personal and human level, for individuals it helps them learn how to intervene in more or less familiar situations that are far removed from the usual ones. Swiss researcher Philippe Perrenoud describes savoir-agir as “the ability to act effectively in a defined type of situation”. This means effectively mobilizing a set of resources (theoretical knowledge, coping strategies, etc.) in a real-life situation. OAK Tutor helps operators to develop action skills corresponding to the specific needs of the company and the immediate situation.