Abstract:
The paper proposes a technique for user-friendly, customizable analytical text processing. The technique is computationally efficient and does not require in-depth user knowledge of formal pattern specification languages. The approach is targeted at recognition of relatively simple contexts, which are unknown beforehand, when it is impractical to develop a full-scale rule-based system for recognition of textual situations. High efficiency is based on document collection preprocessing and preliminary filtering.
Key words and phrases:natural language processing, pattern matching, context filtering, finite automata.
Fulltext:
PDF file (494 kB)