Dictionary Browser Cut Board Info Help English | Hungarian

Automatically Generated Online Dictionaries

Objective

• To find a method that facilitates the creation of bilingual dictionaries by automatic means .
• Why? To help the production of dictionaries for lesser used languages by reducing the human effort needed for lexicographic work.

Method

• Let's take parallel corpora as our starting point and use a statistical method ( word alignment ) to find the optimal pairings of source words and target words !

• Parallel corpus:

  

• Word alignment:

Pros

• Time and cost-effective

  • It might be worth investing in the production of dictionaries for lesser used language pairs.

• Decreases the reliance on human intuition during lexicographic work

  • The method focuses on the actual language usage of several speakers.
  • The dictionary building starts from representative, observable linguistic data, both in the source and in the target language.
    

• Give hints about the best translation candidate
  

  • Provides great amount of natural language contexts
  • Translation candidates can be ranked by their translation probabilities: most frequently used candidates come first within an entry.

Cons

• The creation of parallel corpora is rather tedious for lesser used languages.

• The method in its present state deals only with one-token units .

• The resulting dictionary might include incorrect translation candidates , as well.

  • The precision of the results can be increased by introducing statistical filters.
    
  • Manual post-editing is needed to produce a "traditional" dictionary from the automatically generated database.
    

Workflow (Hungarian-Lithuanian)