Automatic Acquisition of Possible Contexts for Low-Frequent Words
Silvia Necşulescu · 2011
The present work constitutes a PhD project that aims to overcome the problem caused by data sparsity in the task of acquisition of lexical resources. In any corpus of any length, many words are infrequent, thus they co-occur with a small set of words. Nevertheless, they can co-occur with many other words. Our goal is to discover some more possible co-occurring words for low-frequent words relying on other co-occurrences observed in corpus. Our approach aims to formulate a new similarity measure, based on the words usage in language, to approve a transfer of co-occurring words, from a frequent word to a low-frequent word. 1.