Sentiment Analysis for Learning Analytics

AI-supported sentiment analysis applied within the K-12 educational sector is poised to redefine how we understand what students understand. Until now, educators have been relegated to an interpretation of sentiment made on assumption. We infer sentiment based on what is communicated from the student, which can easily be skewed by a student unable to articulate their meta-cognition or a student aiming to say or emote what they think is desirable by their teacher.

When we think about learning analytics, we think about numbers on a page or dots on a graph. Educators try to infuse intrapersonal data to complete the circle of the whole student, but even the most well-intentioned teacher evaluation is susceptible to human error. Sentiment analysis can quantify human emotion, which can then be cross-referenced with academic data. Essentially, it is another dimension of data that has only been valued recently in the k-12 space. How we feel about what we are learning, or how we are learning, or how we are being assessed can give powerful traction to connecting learners to a personalized learning system based on predictive and prescriptive data analysis powered by real multi-dimensional data.

With this type of in-depth evaluation, adjustments to curriculum, teacher support, and professional development can be done with a more accurate picture of the efficacy of current resources. However, this type of data collection cannot be a one-and-done endeavor. If k-12 entities are going to use sentiment analysis, it has to be understood that there must be a feedback loop for continuous improvement. If ever there were data likely to ebb and flow and change radically, it is that of the human emotion. Data sets can become obsolete at any moment. All it takes is a new technology to sweep the nation and put information at the fingertips of students and teachers in a way that has never happened before… but, how often does that happen?