Lexical features (unigrams, bigrams, co-occurrences, and target
co-occurrences) are clustered in the LSA methodology based on the
contexts in which they occur.
This relies on a feature by context representation of the data, which indicates the contexts in which a given feature occurs. This is to be contrasted with the word (unigram) clustering supported by the native SenseClusters methodology, which clusters words based on the words with which they co-occur.
The input must be a Senseval-2 formatted test file. It can be either headed or headless. Even if the data has target words (headed) the test_scope option and target co-occurrence features are not available. A separate set of feature selection data (ie., training data) may not be used with feature clustering.