Translation with memory (censhare Client)

Introduction

Translation of Content with Memory needs an additional license, which is not included in the standard price.

 The window of "Translation of Content with Memory" is similar to the representation of the "Translation of Content." but it shows translation suggestions based on the global translation memory. With long texts or a large translation memory there will be a pause until the translation memory is loaded and compared with each segment of the document to generate a suggestion list of possible matches. Initially, the server reports the results and then the client compares the suggestions with the content and sorts the list by match factor.
Once you have selected a segment in the source or the target content, you can start translating. If you want to take over a translation from the result list, this can be done by double-clicking the entry or pressing "Apply". Now the translation suggestion can be further edited. Clicking on the thumb icon or the green checkmark (censhare 4.3.x) stores the translation and moves on to the next untranslated text segment.
Already translated segments are currently not proposed automatically if they occur twice or more in an asset. Only after the text has been saved and reopened with "Translation of content with memory" any additional occurrence is recognized as a 100% match and becomes translated and filled in automatically. It is possible to set a threshold in the client settings for which the translation is automatically applied (for example, 80 or 0.8.) Those 80 percent could mean that in a sentence of five words, there are four matching words in the translation memory and one has yet to be translated.
However, no distinction is made between word pairs, phrases, or word order. So this percentage can be much lower. This is because, for analysis and calculation of the match, a reduction of the word to its root is taken into account (stemming). Stop words like (the, and, or) are not considered in order to not inflate the database with those words. These are words that necessarily occur in almost every sentence but do not further describe the content. Conjugated verbs like "drive" and "driven" count as a similar result in word analysis.
To specify the matching, the translation memory uses a separate CDB index for 100% matches. Similar matches are determined using a fuzzy search. Elements that are identical in source and target language and do not need to be retyped, are represented in the segment selector by placeholder tags (for example digits).
You might also want to click on a suggestion with a lower percentage and by clicking on "Apply" use it as a basis for your further translation. Right-click in both translation modules reveals the context menu with a set of predefined special characters and line breaks or your own defined Unicode characters (as ISO 8859-1 Latin 1). These can be inserted directly into the translation without being added to the translation memory.

Terminology

Via the menu command "Window · Translation/Terminology" you open the application of the same name. Here you can search the translation memory. You can also manage translation memory and eliminate or edit invalid entries or manually add new translations. Double-clicking on an item, lets you correct an existing item in your translation memory. Editing of entries is sometimes also necessary after you have imported TMX files from other programs. In the pop-up menu, you can switch between translation memory and terminology. Terminology replaces the dictionary of earlier censhare versions.