Background This paper presents a novel approach to the problem of hedge detection, which involves identifying so-called hedge cues for labeling sentences as certain or uncertain. This is the classification problem for Task 1 of the CoNLL-2010 Shared Task, which focuses on hedging in the biomedical domain. We here propose to view hedge detection as a simple disambiguation problem, restricted to words that have previously been observed as hedge cues. As the feature space for the classifier is still very large, we also perform experiments with dimensionality reduction using the method of random indexing. Results The SVM-based classifiers developed in this paper achieves the best published results so far for sentence-level uncertainty prediction on the CoNLL-2010 Shared Task test data. We also show that the technique of random indexing can be successfully applied for reducing the dimensionality of the original feature space by several orders of magnitude, without sacrificing classifier performance. Conclusions This paper introduces a simplified approach to detecting speculation or uncertainty in text, focusing on the biomedical domain. Evaluated at the sentence-level, our SVM-based classifiers achieve the best published results so far. We also show that the feature space can be aggressively compressed using random indexing while still maintaining comparable classifier performance.