Multilingual semantic lexicon
From Digital Grammars Wiki
Dictionary lookup is an important step in translating from one language to another. Typically, a lookup gives many alternative answers, and choosing the correct one may be a difficult issue. For instance, the Swedish word "val" gives three English equivalents: "choice", "election", and "whale". An adequate support tool for dictionary lookup must notice these alternatives, rather than just silently choose one of them. Ideally, the choice is semantic: the chosen translation is the one that expresses the same meaning as the original.
In the proposed project, the goal is to build an electronically usable semantic lexicon. The lexicon will contain words in different languages, and all relevant mappings between words. The lexicon must be usable for automatic translation in GF (http://gf.digitalgrammars.com), and it must therefore give all grammatical information needed for this. For instance, the Swedish word "val" must be represented as being a noun, and its two grammatical behaviours must be distinguished ("ett val, flera val" vs. "en val, flera valar"). The GF Resource Grammar Library gives support for implementing this lexicon in an efficient way.
In the first MSc thesis in the project, it is enough build a Swedish lexicon corresponding to an English lexicon of 6,000 words that is given. The work should be automatized as much as possible, by using resources such as Lexin, WordNet, Wiktionary, and Google Translation. Designing and evaluating automatic lexicon building techniques is a part of the project as much as the final lexicon itself. Later MSc theses can add languages to the system, and also extend the lexicon with new words, to technical vocabularies, etc.
