Interestingness

Interestingness is a quality of information, data, or patterns that capture attention, curiosity, or perceived relevance. It is used in information science, data mining, machine learning, and human-computer interaction to explain why some results are more likely to be explored or acted upon than others. The notion blends novelty, surprise, usefulness, and coherence with existing knowledge.

Because interest is partly subjective, assessments of interestingness combine objective criteria with user models and context. In computational settings, patterns or results are ranked by measures intended to predict usefulness or appeal rather than mere existence. A family of interestingness measures in data mining includes lift, leverage, conviction, and novelty, among others.

In information retrieval and pattern discovery, interestingness guides ranking and filtering, helping to reduce information overload and highlight valuable insights. Theoretical frameworks draw on information theory (surprisal, entropy, information gain) and probabilistic modeling (Bayesian surprise, utility-based approaches) to quantify how unexpected or informative a result is likely to be for a user.

Applications span search engines, recommender systems, dashboards, and exploratory data analysis. Critics note that any single metric can bias results or overemphasize novelty at the expense of usefulness. Effective systems often tailor interestingness to user goals, domain constraints, and evolving knowledge.

See also: curiosity, novelty, relevance, surprise, information theory, data mining, recommender systems.