![]() Custom Translator will only match words and phrases in the input sentence that use exactly the same capitalization and punctuation marks as specified in the source dictionary file. Dictionary entries are case- and punctuation-sensitive. If you're using a phrase dictionary, capitalization and punctuation are important. The best practice is to avoid phrase dictionary entries for anything but compound nouns. It doesn't work as well for verbs or adjectives because those words are typically highly contextual within the source or target language. The phrase dictionary works well for compound nouns like product names (" Microsoft SQL Server"), proper names (" City of Hamburg"), or product features (" pivot table"). The result is that, while the phrase or word within the sentence will translate according to the provided dictionary, the overall translation quality of the sentence often suffers. When a phrase within a sentence is replaced, the context of that sentence is lost or limited for translating the rest of the sentence. The phrase dictionary should be used sparingly. However, when sentences or compound nouns must be translated verbatim, use a dictionary. For better results, we recommended letting the system learn from your training data. Recommendationsĭictionaries aren't a substitute for training a model using training data. You won't get a test report.Ĭustom Translator doesn't sentence align dictionary files, so it is important that there are an equal number of source and target phrases/sentences in your dictionary documents and that they are precisely aligned. The resulting models will use the Microsoft baseline models for translation with the addition of the dictionaries you've added. Your model will typically complete training much faster than a standard training. Since this training is dictionary-only, there's no minimum number of training sentences required. To do so, select only the dictionary document (or multiple dictionary documents) that you wish to include and select Create model. You can train a model using only dictionary data. When a match is detected, the target entry of the sentence dictionary will be returned. If only a portion of the sentence matches, the entry won't match. If the source dictionary entry ends with punctuation, it's ignored during the match. For a sentence dictionary match to occur, the entire submitted sentence must match the source dictionary entry. The sentence dictionary allows you to specify an exact target translation for a source sentence. Sentence dictionaryĪ sentence dictionary is case-insensitive. You can use a phrase dictionary to specify phrases that shouldn't be translated by providing the same untranslated phrase in the source and target files. The rest of the sentence is translated as usual. When you include a phrase dictionary in training your model, any word or phrase listed is translated in the way specified. It's an exact find-and-replace operation. Phrase dictionaryĪ phrase dictionary is case-sensitive. However, a customer provided dictionary takes precedent and will be searched first to look up words or sentences.ĭictionaries only work for projects in language pairs that have a fully supported Microsoft general neural network model behind them. Furthermore, Microsoft Custom Translator service builds and makes use of its own general purpose dictionaries to improve the quality of its translation. You can think of the dictionary as a brute force "copy and replace" for all the terms you list. ![]() Dictionaries are sometimes called glossaries or term bases. ![]() Use a dictionary in your training, when you want Translator to translate any instances of the source phrase or sentence, using the translation you've provided in the dictionary. A dictionary is an aligned pair of documents that specifies a list of phrases or sentences and their corresponding translations. ![]()
0 Comments
Leave a Reply. |