Results

The methodology presented in this paper are the basis for building a Controlled Vocabulary of Concepts (CVC) for the support of Deep Semantic Search. The elements of the Foundation Core Glossary (FCG) were carefully selected from the Minimal English Lexicon (MEL) [GoWi2015a], Longman's Defining Vocabulary [Long2014] with about 2,000 words, and the noun categories of Wordnet. The experience gained from modeling has shown that even for rather complex Semantic Compounds (SC), the maxLevelof CBT definitions tends not to be larger than 4 to 6. The leafes of the Concept Binary Trees as depicted in Figure wsd-barometer2 are always semantic primes, e.g. air, pressure, aid, thing, and measurement. In the approach presented here, each concept is bootstrapped from English and supplemented with translations from the other languages. Consequently, it is not necessary to link concepts between languages via SynSets. Instead, concepts are already linked at the deepest glossary level. This saves an enormous amount of data and work. Compared to WordNet Search, SbM allows to search for arbitrary ontological sub-concepts. Therefore, SbM can be a valuable tool for editors to check if concepts are already included in the CVC before inserting a new concept. Other lexicons and thesauri enrich concepts and their meanings mainly by extra-concept relations to other concepts. With intra-concept relations (ICRs), we have added new modeling possibilities. To the best of our knowledge, there is no comparable approach. This paper shows that Search by Meaning (SbM) goes beyond traditional search by systematically considering the internal ontological and linguistic semantics of concepts.