en:manualy:treq - Příručka ČNK

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.

====== Treq ======

{{ :manualy:treq.png?direct&200|}}

The [[http://treq.korpus.cz|Treq]] application serves for searching Czech - foreign language dictionaries, which have been automatically created based on data derived from the parallel corpus [[en:cnk:intercorp|InterCorp]]. It enables users to search for all possible equivalents used in translations, or to find synonyms.

Treq is an online application (the only thing we need to use it is a web browser) and it is accessible without [[en:kurz:zaciname|registration]] to all users at **[[http://treq.korpus.cz|treq.korpus.cz]]**.

When searching, it is necessary to first select a foreign language (//Language 2://) and the language of the input word (//Search in://). The word may be typed in a specific [[en:pojmy:word|form]] or as [[en:pojmy:lemma|lemma]] (//Lemma: Yes/No//). We may also choose whether the result should be based on the fiction [[en:cnk:intercorp:verze8#obsah_korpusu|core]], individual collections or all texts in the given version of the [[en:cnk:intercorp|InterCorp]] corpus with the help of (//Restrict to://). We input the word into the (//Query://) field and click on //Search//. 

The result of the query is a list of found translations of the given word, listed based on frequency. To check the occurrences in their context, the pair can be searched for in the [[en:manualy:kontext:index|KonText]] interface with a query in the InterCorp corpus, which we can initiate by clicking on the selected equivalent. However, the number of occurrences can differ – a parallel query will find concordances in which the potential equivalent is a different word.

===== Alignment principle =====

The original and translated texts are firstly (based on statistical calculations) aligned word by word with the help of the program GIZA++((see [[http://dx.doi.org/10.1162/089120103321337421|Och–Ney 2003]])). The aligned pairs of words are then sorted and summarized. The results of automatic extraction are not revised in any way, but the relative frequency of the given pair of words can serve as an indicator of relevance. The more frequent the occurrence of the equivalent in comparison with the other equivalents, the higher the probability that is is functional. 

===== Application pictures =====

[{{:manualy:treq-form.png?direct&300|Input form}}]
[{{:manualy:treq-skorapka.png?direct&300|Searching in the Czech-English section}}]
[{{:manualy:treq-warum.png?direct&300|Searching in the Czech-German section}}]

==== Related links ====

<WRAP round box 50%>
[[en:manualy:kontext:index|KonText interface]] • [[syd|SyD]] • [[morfio|Morfio]] • [[kwords|KWords]] • [[en:pojmy:korpusovy_manazer|Corpus Manager]] • [[en:pojmy:nastroje|Corpus tools]]
</WRAP>

Trace: • pokrocile_dotazy • ipm • akalex • gen2 • mi-score • klasifikace_textu_syn2015 • korpusy • kh-noviny • asociacni_miry • morfologicka_analyza