摘要：AbstractWe propose a symbiotic framework in which correspondences between electronic multilingual lexicons and translation example banks can be captured, so that their functions and contents may benefit and improve upon one another. Several mechanisms are used for this purpose: i.) two flexible annotation schemas, S-SSTC and SSTC+L, for supporting irregular multi-level correspondences across languages; ii.) an axis-based translation cluster structure for connecting translation equivalents; and iii.) translation profiles, for capturing contexts of translation equivalent instances in the corpora. There are two main contributions: i.) the design of SSTC+L, which allows the annotation of multi-word expressions and transla- tion lexical gaps in the translation examples; and ii.) the overall framework facilitating the symbiotic flow of information between the multilingual lexicon and translation bank. We give illustrative examples to show how these mechanisms can be used for translation selection, addition of new language items, and verification of lexicon contents. Preliminary tests show there is potential in our approach.
