A Computational Approach to Generate a Sensorial Lexicon

While humans are capable of building connections between words and sensorial modalities by using commonsense knowledge, it is not straightforward for machines to interpret sensorial information. To this end, a lexicon associating words with human senses, namely sight, hearing, taste, smell and touch, would be crucial. Nonetheless, to the best of our knowledge, there is no systematic attempt in the literature to build such a resource. In this paper, we propose a computational method based on bootstrapping and corpus statistics to automatically associate English words with senses. To evaluate the quality of the resulting lexicon, we create a gold standard via crowdsourcing and show that a simple classifier relying on the lexicon outperforms two baselines on a sensory classification task, both at word and sentence level. The results confirm the soundness of the proposed approach for the construction of the lexicon and the usefulness of the resource for computational applications.