Automatic Acquisition of Domain Specific Lexicons

In this paper we present the results of three years of experiments about automatic acquisition of domain specific terminology from corpora. We present an analysis of the potentiality and limitations of the Term Categorization approach to lexical acquisition, and we propose a novel methodology to approach the task, consisting on applying Latent Semantic Kernels to estimate term similarity. We find out that domain specific monosemous terms behave similarly to domain specific lexical items, so we used them to train and evaluate our Term Categorization system. Results show that the proposed technique is effective, achieving an accuracy of about 43%. We also reported an error analysis showing that most of the misclassification errors are related the the fuzzy nature of domain distinctions. In particular we identified a set of “families” in the WordNet Domains categories that makes difficult the classification task. Categorizing monosemous terms according to domain labels allows us to automatically assigning domain labels to a subset of the WordNet synsets, allowing to perform a bootstrap procedure to assign a of domain label to every synset in WordNet.