Some Ways of Determining Word Combinations for Classifying Texts

摘要

Different ways of defining the term "word combinations" are considered. New methods based on syntactic trees are examined, as is the use of n-grams. Terms such as n-chain, n-subtree, and full t-subtrees are introduced. New types of word combinations can serve as the basis for start-of-text characters in problems of computer-assisted teaching. All described types of word combinations are compared to one another in a task of determining authorship, posed as a classification problem.