Bridging Artificial Intelligence and Information Behavior: A Multidisciplinary Study of User Information Need Prediction

Abstract

The increased overlap between artificial intelligence (AI), data science and information behaviour study offers radical possibilities in how users expect and understand intentions in the digital ecosystems. This paper presents a multidisciplinary analytical framework which combines behavioural analytics, linguistic modelling, and machine learning and predicts user information demands with high accuracy. One hundred and fifty thousand search sessions of the many educational repositories were examined to determine the cognitive and language patterns affecting search-changeover and decision progression. Random Forest and Long Short-Term Memory (LSTM) networks were able to perform well through intensive feature derivation and model optimization to achieve more than 92% predictive accuracy. The findings substantiate a close interaction of human cognitive behaviour and algorithmic inference, and show the way the latent user intent is effectively represented by computational models. Along with the performance indicators, the present study highlights the multi-domain opportunities of AI to equip the digital world with further literacy rates, improve the accessibility of information in equal terms, and assist smart decision-making at the educational and professional levels. The offered framework will be the basis of human-oriented AI systems which learn as they observe behavioural patterns, which will result in the next step of context-sensitive information retrieval and personalised knowledge discovery.